{"qa_id": "DAVIS__blackswan__object_world_motion_direction__00", "video_id": "DAVIS__blackswan", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (black swan with a red beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "backward", "right", "up"], "frame_paths": ["media/12600f0e31209c7c75f2/F0.jpg", "media/12600f0e31209c7c75f2/F1.jpg", "media/12600f0e31209c7c75f2/F2.jpg", "media/12600f0e31209c7c75f2/F3.jpg", "media/12600f0e31209c7c75f2/F4.jpg", "media/12600f0e31209c7c75f2/F5.jpg", "media/12600f0e31209c7c75f2/F6.jpg", "media/12600f0e31209c7c75f2/F7.jpg", "media/12600f0e31209c7c75f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (black swan with a red beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b29536f5c875a4486e3d9f7a7b49492e80cd65102998919e7e08505f2b6694fa"}
{"qa_id": "DAVIS__bmx-bumps__camera_view_direction_change__00", "video_id": "DAVIS__bmx-bumps", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/b3a893f76d8c4b879a23/F0.jpg", "media/b3a893f76d8c4b879a23/F1.jpg", "media/b3a893f76d8c4b879a23/F2.jpg", "media/b3a893f76d8c4b879a23/F3.jpg", "media/b3a893f76d8c4b879a23/F4.jpg", "media/b3a893f76d8c4b879a23/F5.jpg", "media/b3a893f76d8c4b879a23/F6.jpg", "media/b3a893f76d8c4b879a23/F7.jpg", "media/b3a893f76d8c4b879a23/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "51168c2068a29e100b579d91d722b1e48b02cffec0e47a768ba54427444aee08"}
{"qa_id": "DAVIS__bmx-bumps__object_world_motion_direction__01", "video_id": "DAVIS__bmx-bumps", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (cyclist in black and red gear with number 91) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/98d39d880a7cc48a3a76/F0.jpg", "media/98d39d880a7cc48a3a76/F1.jpg", "media/98d39d880a7cc48a3a76/F2.jpg", "media/98d39d880a7cc48a3a76/F3.jpg", "media/98d39d880a7cc48a3a76/F4.jpg", "media/98d39d880a7cc48a3a76/F5.jpg", "media/98d39d880a7cc48a3a76/F6.jpg", "media/98d39d880a7cc48a3a76/F7.jpg", "media/98d39d880a7cc48a3a76/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (cyclist in black and red gear with number 91) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "766ece6334c8201b4e745b2eef82b4f10718e3c8a7c62462cb9a5e7057b88d01"}
{"qa_id": "DAVIS__car-roundabout__camera_view_direction_change__00", "video_id": "DAVIS__car-roundabout", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/d12c57c1c12e79eb4045/F0.jpg", "media/d12c57c1c12e79eb4045/F1.jpg", "media/d12c57c1c12e79eb4045/F2.jpg", "media/d12c57c1c12e79eb4045/F3.jpg", "media/d12c57c1c12e79eb4045/F4.jpg", "media/d12c57c1c12e79eb4045/F5.jpg", "media/d12c57c1c12e79eb4045/F6.jpg", "media/d12c57c1c12e79eb4045/F7.jpg", "media/d12c57c1c12e79eb4045/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55c282928f8940355d2075dba86d8619c91a164d28fb231d532d13fe514a5de1"}
{"qa_id": "DAVIS__car-roundabout__object_world_motion_direction__01", "video_id": "DAVIS__car-roundabout", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (dark grey mini cooper) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "forward", "up"], "frame_paths": ["media/d8f2d7c08c23e257a8ae/F0.jpg", "media/d8f2d7c08c23e257a8ae/F1.jpg", "media/d8f2d7c08c23e257a8ae/F2.jpg", "media/d8f2d7c08c23e257a8ae/F3.jpg", "media/d8f2d7c08c23e257a8ae/F4.jpg", "media/d8f2d7c08c23e257a8ae/F5.jpg", "media/d8f2d7c08c23e257a8ae/F6.jpg", "media/d8f2d7c08c23e257a8ae/F7.jpg", "media/d8f2d7c08c23e257a8ae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (dark grey mini cooper) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bbc2eefc020c5e27ae214008cb7fb36e0df1b8bf5990184d3a9cf3d27ca1585a"}
{"qa_id": "DAVIS__car-shadow__camera_view_direction_change__00", "video_id": "DAVIS__car-shadow", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/3657eef25ef0d1d81b83/F0.jpg", "media/3657eef25ef0d1d81b83/F1.jpg", "media/3657eef25ef0d1d81b83/F2.jpg", "media/3657eef25ef0d1d81b83/F3.jpg", "media/3657eef25ef0d1d81b83/F4.jpg", "media/3657eef25ef0d1d81b83/F5.jpg", "media/3657eef25ef0d1d81b83/F6.jpg", "media/3657eef25ef0d1d81b83/F7.jpg", "media/3657eef25ef0d1d81b83/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "DAVIS__car-shadow__object_world_motion_direction__01", "video_id": "DAVIS__car-shadow", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (silver compact car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "up", "forward", "down"], "frame_paths": ["media/cf680887ddf854190449/F0.jpg", "media/cf680887ddf854190449/F1.jpg", "media/cf680887ddf854190449/F2.jpg", "media/cf680887ddf854190449/F3.jpg", "media/cf680887ddf854190449/F4.jpg", "media/cf680887ddf854190449/F5.jpg", "media/cf680887ddf854190449/F6.jpg", "media/cf680887ddf854190449/F7.jpg", "media/cf680887ddf854190449/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (silver compact car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f851959dbecee326ffdff6805525901a6dfcc332e87aec2a5c3a054c79213ba6"}
{"qa_id": "DAVIS__car-shadow__camera_object_distance_change__02", "video_id": "DAVIS__car-shadow", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (silver compact car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/cf680887ddf854190449/F0.jpg", "media/cf680887ddf854190449/F1.jpg", "media/cf680887ddf854190449/F2.jpg", "media/cf680887ddf854190449/F3.jpg", "media/cf680887ddf854190449/F4.jpg", "media/cf680887ddf854190449/F5.jpg", "media/cf680887ddf854190449/F6.jpg", "media/cf680887ddf854190449/F7.jpg", "media/cf680887ddf854190449/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (silver compact car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31e12813560ba3e069993cc2bc9eedfcf782a62f6fd5245dc682cd29dca470d5"}
{"qa_id": "DAVIS__car-turn__camera_view_direction_change__00", "video_id": "DAVIS__car-turn", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/a64b49424d4941175a6d/F0.jpg", "media/a64b49424d4941175a6d/F1.jpg", "media/a64b49424d4941175a6d/F2.jpg", "media/a64b49424d4941175a6d/F3.jpg", "media/a64b49424d4941175a6d/F4.jpg", "media/a64b49424d4941175a6d/F5.jpg", "media/a64b49424d4941175a6d/F6.jpg", "media/a64b49424d4941175a6d/F7.jpg", "media/a64b49424d4941175a6d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64e0fec494c815ab1788b70a6ff118230da5d701e043aa379365e44f1a46e68d"}
{"qa_id": "DAVIS__car-turn__camera_object_distance_change__01", "video_id": "DAVIS__car-turn", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (silver compact SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b93c447335a0bf95b1a5/F0.jpg", "media/b93c447335a0bf95b1a5/F1.jpg", "media/b93c447335a0bf95b1a5/F2.jpg", "media/b93c447335a0bf95b1a5/F3.jpg", "media/b93c447335a0bf95b1a5/F4.jpg", "media/b93c447335a0bf95b1a5/F5.jpg", "media/b93c447335a0bf95b1a5/F6.jpg", "media/b93c447335a0bf95b1a5/F7.jpg", "media/b93c447335a0bf95b1a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (silver compact SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1e1349a78034f745464d6e1bbac63b21cbc82dce16838ef7ee1f00cddca24904"}
{"qa_id": "DAVIS__crossing__object_pair_distance_change__00", "video_id": "DAVIS__crossing", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (person wearing a grey t-shirt and orange shorts) and object B (white delivery van with green text on the side) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/029b05a026067d69ede2/F0.jpg", "media/029b05a026067d69ede2/F1.jpg", "media/029b05a026067d69ede2/F2.jpg", "media/029b05a026067d69ede2/F3.jpg", "media/029b05a026067d69ede2/F4.jpg", "media/029b05a026067d69ede2/F5.jpg", "media/029b05a026067d69ede2/F6.jpg", "media/029b05a026067d69ede2/F7.jpg", "media/029b05a026067d69ede2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person wearing a grey t-shirt and orange shorts) and object B (white delivery van with green text on the side) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e07f0b9fcd67d9274e4d3c4322b64a8bc3d1cd9705cd68861d1301bc9c23e605"}
{"qa_id": "DAVIS__dog__camera_translation_direction__00", "video_id": "DAVIS__dog", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "forward", "backward", "left"], "frame_paths": ["media/27710d980514319c80a8/F0.jpg", "media/27710d980514319c80a8/F1.jpg", "media/27710d980514319c80a8/F2.jpg", "media/27710d980514319c80a8/F3.jpg", "media/27710d980514319c80a8/F4.jpg", "media/27710d980514319c80a8/F5.jpg", "media/27710d980514319c80a8/F6.jpg", "media/27710d980514319c80a8/F7.jpg", "media/27710d980514319c80a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "86306f6410a73a3e4aa14c28ed57b420631bc059e8a38ee91711291b0aa01dd9"}
{"qa_id": "DAVIS__dog__camera_view_direction_change__01", "video_id": "DAVIS__dog", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/27710d980514319c80a8/F0.jpg", "media/27710d980514319c80a8/F1.jpg", "media/27710d980514319c80a8/F2.jpg", "media/27710d980514319c80a8/F3.jpg", "media/27710d980514319c80a8/F4.jpg", "media/27710d980514319c80a8/F5.jpg", "media/27710d980514319c80a8/F6.jpg", "media/27710d980514319c80a8/F7.jpg", "media/27710d980514319c80a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654016ae105affb53bc2face594c67e96f587f4cdc806a6f418076b30032d255"}
{"qa_id": "DAVIS__dog__object_world_motion_direction__02", "video_id": "DAVIS__dog", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object A (white dog with a collar) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "right", "backward"], "frame_paths": ["media/2c16ab7a8f555b187bf8/F0.jpg", "media/2c16ab7a8f555b187bf8/F1.jpg", "media/2c16ab7a8f555b187bf8/F2.jpg", "media/2c16ab7a8f555b187bf8/F3.jpg", "media/2c16ab7a8f555b187bf8/F4.jpg", "media/2c16ab7a8f555b187bf8/F5.jpg", "media/2c16ab7a8f555b187bf8/F6.jpg", "media/2c16ab7a8f555b187bf8/F7.jpg", "media/2c16ab7a8f555b187bf8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object A (white dog with a collar) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0db3fcc316ae9298e258db92c7cb29e73acabcd07bcb4c9765989ad76e2ff95d"}
{"qa_id": "DAVIS__dog__camera_object_distance_change__03", "video_id": "DAVIS__dog", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (white dog with a collar) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2c16ab7a8f555b187bf8/F0.jpg", "media/2c16ab7a8f555b187bf8/F1.jpg", "media/2c16ab7a8f555b187bf8/F2.jpg", "media/2c16ab7a8f555b187bf8/F3.jpg", "media/2c16ab7a8f555b187bf8/F4.jpg", "media/2c16ab7a8f555b187bf8/F5.jpg", "media/2c16ab7a8f555b187bf8/F6.jpg", "media/2c16ab7a8f555b187bf8/F7.jpg", "media/2c16ab7a8f555b187bf8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (white dog with a collar) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e655a6668f357ec6e99732c0eefe212748716856dc516845c31fec98cfde958"}
{"qa_id": "DAVIS__dogs-scale__object_world_motion_direction__00", "video_id": "DAVIS__dogs-scale", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (black and tan puppy standing on a white scale) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "backward", "up", "forward"], "frame_paths": ["media/332bff6630cc00ab59b0/F0.jpg", "media/332bff6630cc00ab59b0/F1.jpg", "media/332bff6630cc00ab59b0/F2.jpg", "media/332bff6630cc00ab59b0/F3.jpg", "media/332bff6630cc00ab59b0/F4.jpg", "media/332bff6630cc00ab59b0/F5.jpg", "media/332bff6630cc00ab59b0/F6.jpg", "media/332bff6630cc00ab59b0/F7.jpg", "media/332bff6630cc00ab59b0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (black and tan puppy standing on a white scale) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05adeeeb8c104e74055ea662874fe262706b0aa8aa518034df466904755bd44e"}
{"qa_id": "DAVIS__dogs-scale__object_travel_distance_comparison__01", "video_id": "DAVIS__dogs-scale", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black and tan puppy standing on a white scale) or B (black and tan puppy sitting on the wooden table)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/287014951ece717d7111/F0.jpg", "media/287014951ece717d7111/F1.jpg", "media/287014951ece717d7111/F2.jpg", "media/287014951ece717d7111/F3.jpg", "media/287014951ece717d7111/F4.jpg", "media/287014951ece717d7111/F5.jpg", "media/287014951ece717d7111/F6.jpg", "media/287014951ece717d7111/F7.jpg", "media/287014951ece717d7111/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black and tan puppy standing on a white scale) or B (black and tan puppy sitting on the wooden table)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d53dcd8b6f4a1112d45eb20c9fac3239cb31776a90c0e3c40ff3c249d8af38c"}
{"qa_id": "DAVIS__drift-turn__camera_object_distance_change__00", "video_id": "DAVIS__drift-turn", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did object A (white and orange drift car with large rear wing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2dc0d9002267782b13f9/F0.jpg", "media/2dc0d9002267782b13f9/F1.jpg", "media/2dc0d9002267782b13f9/F2.jpg", "media/2dc0d9002267782b13f9/F3.jpg", "media/2dc0d9002267782b13f9/F4.jpg", "media/2dc0d9002267782b13f9/F5.jpg", "media/2dc0d9002267782b13f9/F6.jpg", "media/2dc0d9002267782b13f9/F7.jpg", "media/2dc0d9002267782b13f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (white and orange drift car with large rear wing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7347c322dd58da1a39a21e23752fd4b92d984df6b5c482273fa795f58bf4e8d2"}
{"qa_id": "DAVIS__flamingo__object_world_motion_direction__00", "video_id": "DAVIS__flamingo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (white flamingo with pink-tipped wings and long orange legs wading in water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "forward", "right", "left"], "frame_paths": ["media/4ba9d1ea52f19716df4b/F0.jpg", "media/4ba9d1ea52f19716df4b/F1.jpg", "media/4ba9d1ea52f19716df4b/F2.jpg", "media/4ba9d1ea52f19716df4b/F3.jpg", "media/4ba9d1ea52f19716df4b/F4.jpg", "media/4ba9d1ea52f19716df4b/F5.jpg", "media/4ba9d1ea52f19716df4b/F6.jpg", "media/4ba9d1ea52f19716df4b/F7.jpg", "media/4ba9d1ea52f19716df4b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (white flamingo with pink-tipped wings and long orange legs wading in water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0d13ed8aa7f9b8fcc78f5bbda6ccf0b04a066a669d40775013628e43bc3fb59"}
{"qa_id": "DAVIS__gold-fish__object_world_motion_direction__00", "video_id": "DAVIS__gold-fish", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange fish with flowing fins) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "down", "left", "up"], "frame_paths": ["media/b9ac396e6bdbedb7d95d/F0.jpg", "media/b9ac396e6bdbedb7d95d/F1.jpg", "media/b9ac396e6bdbedb7d95d/F2.jpg", "media/b9ac396e6bdbedb7d95d/F3.jpg", "media/b9ac396e6bdbedb7d95d/F4.jpg", "media/b9ac396e6bdbedb7d95d/F5.jpg", "media/b9ac396e6bdbedb7d95d/F6.jpg", "media/b9ac396e6bdbedb7d95d/F7.jpg", "media/b9ac396e6bdbedb7d95d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange fish with flowing fins) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5209cedf2e192f95c545306bb65dc991775133998844e8ad791ca09df0673dde"}
{"qa_id": "DAVIS__horsejump-high__camera_object_distance_change__00", "video_id": "DAVIS__horsejump-high", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (equestrian rider in white breeches and orange helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b3fc82532582a73ff0f5/F0.jpg", "media/b3fc82532582a73ff0f5/F1.jpg", "media/b3fc82532582a73ff0f5/F2.jpg", "media/b3fc82532582a73ff0f5/F3.jpg", "media/b3fc82532582a73ff0f5/F4.jpg", "media/b3fc82532582a73ff0f5/F5.jpg", "media/b3fc82532582a73ff0f5/F6.jpg", "media/b3fc82532582a73ff0f5/F7.jpg", "media/b3fc82532582a73ff0f5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (equestrian rider in white breeches and orange helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c17e86f6bd3bb73f6b0bcb54f37229d17fcd941bde6424dddc66f019bc183438"}
{"qa_id": "DAVIS__horsejump-high__camera_object_distance_change__01", "video_id": "DAVIS__horsejump-high", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object B (brown horse jumping over an obstacle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/0c99a728dbb6f5ce3935/F0.jpg", "media/0c99a728dbb6f5ce3935/F1.jpg", "media/0c99a728dbb6f5ce3935/F2.jpg", "media/0c99a728dbb6f5ce3935/F3.jpg", "media/0c99a728dbb6f5ce3935/F4.jpg", "media/0c99a728dbb6f5ce3935/F5.jpg", "media/0c99a728dbb6f5ce3935/F6.jpg", "media/0c99a728dbb6f5ce3935/F7.jpg", "media/0c99a728dbb6f5ce3935/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (brown horse jumping over an obstacle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4a05b885f8844707cb76a8f9f9b3fe9a994a38fa90c5494e12d47955516b1697"}
{"qa_id": "DAVIS__horsejump-low__object_world_motion_direction__00", "video_id": "DAVIS__horsejump-low", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (white horse jumping over an obstacle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "right", "backward", "up"], "frame_paths": ["media/4a441122e9dff2aa33d3/F0.jpg", "media/4a441122e9dff2aa33d3/F1.jpg", "media/4a441122e9dff2aa33d3/F2.jpg", "media/4a441122e9dff2aa33d3/F3.jpg", "media/4a441122e9dff2aa33d3/F4.jpg", "media/4a441122e9dff2aa33d3/F5.jpg", "media/4a441122e9dff2aa33d3/F6.jpg", "media/4a441122e9dff2aa33d3/F7.jpg", "media/4a441122e9dff2aa33d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (white horse jumping over an obstacle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34e5a9a6d143823d3efd5392b6c647dbe1a51318d18dfbc96406415422e50b04"}
{"qa_id": "DAVIS__horsejump-low__object_world_motion_direction__01", "video_id": "DAVIS__horsejump-low", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (equestrian rider in white shirt and black helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/e91b1ee8d661bfd8a71f/F0.jpg", "media/e91b1ee8d661bfd8a71f/F1.jpg", "media/e91b1ee8d661bfd8a71f/F2.jpg", "media/e91b1ee8d661bfd8a71f/F3.jpg", "media/e91b1ee8d661bfd8a71f/F4.jpg", "media/e91b1ee8d661bfd8a71f/F5.jpg", "media/e91b1ee8d661bfd8a71f/F6.jpg", "media/e91b1ee8d661bfd8a71f/F7.jpg", "media/e91b1ee8d661bfd8a71f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (equestrian rider in white shirt and black helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6563b1b8c333a29371f8fd2e8950ad7c9ee366a1a9d680a31c9358b02b6f6d2b"}
{"qa_id": "DAVIS__lab-coat__camera_translation_direction__00", "video_id": "DAVIS__lab-coat", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "right", "down", "backward", "forward"], "frame_paths": ["media/fcbe050cc2b99852642c/F0.jpg", "media/fcbe050cc2b99852642c/F1.jpg", "media/fcbe050cc2b99852642c/F2.jpg", "media/fcbe050cc2b99852642c/F3.jpg", "media/fcbe050cc2b99852642c/F4.jpg", "media/fcbe050cc2b99852642c/F5.jpg", "media/fcbe050cc2b99852642c/F6.jpg", "media/fcbe050cc2b99852642c/F7.jpg", "media/fcbe050cc2b99852642c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. right\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebd153017e229ad8d99f4f55b1a3fe1bef0c22b0c4bc634b2f187e74188a4283"}
{"qa_id": "DAVIS__lab-coat__camera_object_distance_change__01", "video_id": "DAVIS__lab-coat", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, did object A (person with orange hair wearing a white coat and white boots) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/433b5d0437bf103fee04/F0.jpg", "media/433b5d0437bf103fee04/F1.jpg", "media/433b5d0437bf103fee04/F2.jpg", "media/433b5d0437bf103fee04/F3.jpg", "media/433b5d0437bf103fee04/F4.jpg", "media/433b5d0437bf103fee04/F5.jpg", "media/433b5d0437bf103fee04/F6.jpg", "media/433b5d0437bf103fee04/F7.jpg", "media/433b5d0437bf103fee04/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (person with orange hair wearing a white coat and white boots) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "179d60d7484c6b36a8291f220dbf13e4b223cb7c565e0f8016e42f266907c87b"}
{"qa_id": "DAVIS__lab-coat__camera_object_distance_change__02", "video_id": "DAVIS__lab-coat", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, did object B (person with long blonde hair wearing a white coat and dark brown boots) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/964b4b9a33f79bba15c4/F0.jpg", "media/964b4b9a33f79bba15c4/F1.jpg", "media/964b4b9a33f79bba15c4/F2.jpg", "media/964b4b9a33f79bba15c4/F3.jpg", "media/964b4b9a33f79bba15c4/F4.jpg", "media/964b4b9a33f79bba15c4/F5.jpg", "media/964b4b9a33f79bba15c4/F6.jpg", "media/964b4b9a33f79bba15c4/F7.jpg", "media/964b4b9a33f79bba15c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object B (person with long blonde hair wearing a white coat and dark brown boots) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7393e987541ee695e9e20847ef458c662e688cedc350daecd7b7d1b30c8a97a0"}
{"qa_id": "DAVIS__lady-running__camera_translation_direction__00", "video_id": "DAVIS__lady-running", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "backward", "down", "left"], "frame_paths": ["media/3be6f9a8fbe8a92f09a0/F0.jpg", "media/3be6f9a8fbe8a92f09a0/F1.jpg", "media/3be6f9a8fbe8a92f09a0/F2.jpg", "media/3be6f9a8fbe8a92f09a0/F3.jpg", "media/3be6f9a8fbe8a92f09a0/F4.jpg", "media/3be6f9a8fbe8a92f09a0/F5.jpg", "media/3be6f9a8fbe8a92f09a0/F6.jpg", "media/3be6f9a8fbe8a92f09a0/F7.jpg", "media/3be6f9a8fbe8a92f09a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9923fe80779c8e5e81f3eeff9b8cd1b25257f684a5ff33df635919559583118e"}
{"qa_id": "DAVIS__lady-running__camera_view_direction_change__01", "video_id": "DAVIS__lady-running", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/3be6f9a8fbe8a92f09a0/F0.jpg", "media/3be6f9a8fbe8a92f09a0/F1.jpg", "media/3be6f9a8fbe8a92f09a0/F2.jpg", "media/3be6f9a8fbe8a92f09a0/F3.jpg", "media/3be6f9a8fbe8a92f09a0/F4.jpg", "media/3be6f9a8fbe8a92f09a0/F5.jpg", "media/3be6f9a8fbe8a92f09a0/F6.jpg", "media/3be6f9a8fbe8a92f09a0/F7.jpg", "media/3be6f9a8fbe8a92f09a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55c282928f8940355d2075dba86d8619c91a164d28fb231d532d13fe514a5de1"}
{"qa_id": "DAVIS__lady-running__camera_object_distance_change__02", "video_id": "DAVIS__lady-running", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object A (elderly person with short white hair wearing a colorful patterned long-sleeve shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/6d986efe6ec95773c523/F0.jpg", "media/6d986efe6ec95773c523/F1.jpg", "media/6d986efe6ec95773c523/F2.jpg", "media/6d986efe6ec95773c523/F3.jpg", "media/6d986efe6ec95773c523/F4.jpg", "media/6d986efe6ec95773c523/F5.jpg", "media/6d986efe6ec95773c523/F6.jpg", "media/6d986efe6ec95773c523/F7.jpg", "media/6d986efe6ec95773c523/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (elderly person with short white hair wearing a colorful patterned long-sleeve shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3047c634d41801c4ca798204bade3ebc0bea3debc8e7e93c9528742611a88b3b"}
{"qa_id": "DAVIS__libby__camera_translation_direction__00", "video_id": "DAVIS__libby", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "up", "forward", "backward"], "frame_paths": ["media/419785488a7a26cf2c5a/F0.jpg", "media/419785488a7a26cf2c5a/F1.jpg", "media/419785488a7a26cf2c5a/F2.jpg", "media/419785488a7a26cf2c5a/F3.jpg", "media/419785488a7a26cf2c5a/F4.jpg", "media/419785488a7a26cf2c5a/F5.jpg", "media/419785488a7a26cf2c5a/F6.jpg", "media/419785488a7a26cf2c5a/F7.jpg", "media/419785488a7a26cf2c5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a4d29a9bceabaa06eae72e01195b17b644fa6f8d9e288e16fe04b9e993e9101"}
{"qa_id": "DAVIS__libby__camera_view_direction_change__01", "video_id": "DAVIS__libby", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/419785488a7a26cf2c5a/F0.jpg", "media/419785488a7a26cf2c5a/F1.jpg", "media/419785488a7a26cf2c5a/F2.jpg", "media/419785488a7a26cf2c5a/F3.jpg", "media/419785488a7a26cf2c5a/F4.jpg", "media/419785488a7a26cf2c5a/F5.jpg", "media/419785488a7a26cf2c5a/F6.jpg", "media/419785488a7a26cf2c5a/F7.jpg", "media/419785488a7a26cf2c5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c770b4f00ad48393636b1e706c5212e1f6da079e7787d02dd692a1238370dfff"}
{"qa_id": "DAVIS__lindy-hop__object_pair_distance_change__00", "video_id": "DAVIS__lindy-hop", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (man in black shirt and brown pants dancing) and object B (woman in green top and dark shorts dancing) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b28dd16dbf1e2668c0b7/F0.jpg", "media/b28dd16dbf1e2668c0b7/F1.jpg", "media/b28dd16dbf1e2668c0b7/F2.jpg", "media/b28dd16dbf1e2668c0b7/F3.jpg", "media/b28dd16dbf1e2668c0b7/F4.jpg", "media/b28dd16dbf1e2668c0b7/F5.jpg", "media/b28dd16dbf1e2668c0b7/F6.jpg", "media/b28dd16dbf1e2668c0b7/F7.jpg", "media/b28dd16dbf1e2668c0b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (man in black shirt and brown pants dancing) and object B (woman in green top and dark shorts dancing) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4177ab2b789250e5b2f56a5755eaf6aa9a946b9e30443a41faef17930538c3ea"}
{"qa_id": "DAVIS__loading__object_world_motion_direction__00", "video_id": "DAVIS__loading", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object B (blue and black Samsung TV box) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "left", "forward", "right", "backward", "up"], "frame_paths": ["media/a2c28a0d7186215fedfd/F0.jpg", "media/a2c28a0d7186215fedfd/F1.jpg", "media/a2c28a0d7186215fedfd/F2.jpg", "media/a2c28a0d7186215fedfd/F3.jpg", "media/a2c28a0d7186215fedfd/F4.jpg", "media/a2c28a0d7186215fedfd/F5.jpg", "media/a2c28a0d7186215fedfd/F6.jpg", "media/a2c28a0d7186215fedfd/F7.jpg", "media/a2c28a0d7186215fedfd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object B (blue and black Samsung TV box) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. forward\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7795adf6b67c0024b432052a0dc4652abad00f4524aab99e2172f531d3f968e0"}
{"qa_id": "DAVIS__loading__camera_object_distance_change__01", "video_id": "DAVIS__loading", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F6, did object B (blue and black Samsung TV box) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a2c28a0d7186215fedfd/F0.jpg", "media/a2c28a0d7186215fedfd/F1.jpg", "media/a2c28a0d7186215fedfd/F2.jpg", "media/a2c28a0d7186215fedfd/F3.jpg", "media/a2c28a0d7186215fedfd/F4.jpg", "media/a2c28a0d7186215fedfd/F5.jpg", "media/a2c28a0d7186215fedfd/F6.jpg", "media/a2c28a0d7186215fedfd/F7.jpg", "media/a2c28a0d7186215fedfd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object B (blue and black Samsung TV box) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a2a7a58755a3c5a11d30606b793691a6326ab17d01d271c57cfffa60ebb15c5d"}
{"qa_id": "DAVIS__loading__object_pair_distance_change__02", "video_id": "DAVIS__loading", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F6, did the 3D distance between object A (man wearing a blue and grey work jacket) and object B (blue and black Samsung TV box) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/15d567db151e1183a5bf/F0.jpg", "media/15d567db151e1183a5bf/F1.jpg", "media/15d567db151e1183a5bf/F2.jpg", "media/15d567db151e1183a5bf/F3.jpg", "media/15d567db151e1183a5bf/F4.jpg", "media/15d567db151e1183a5bf/F5.jpg", "media/15d567db151e1183a5bf/F6.jpg", "media/15d567db151e1183a5bf/F7.jpg", "media/15d567db151e1183a5bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did the 3D distance between object A (man wearing a blue and grey work jacket) and object B (blue and black Samsung TV box) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8f807ceb44cca68722dfbc93e162d8dee0d1a715e353aafa2b415cf131ba74f"}
{"qa_id": "DAVIS__loading__object_travel_distance_comparison__03", "video_id": "DAVIS__loading", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man wearing a blue and grey work jacket) or B (blue and black Samsung TV box)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/15d567db151e1183a5bf/F0.jpg", "media/15d567db151e1183a5bf/F1.jpg", "media/15d567db151e1183a5bf/F2.jpg", "media/15d567db151e1183a5bf/F3.jpg", "media/15d567db151e1183a5bf/F4.jpg", "media/15d567db151e1183a5bf/F5.jpg", "media/15d567db151e1183a5bf/F6.jpg", "media/15d567db151e1183a5bf/F7.jpg", "media/15d567db151e1183a5bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man wearing a blue and grey work jacket) or B (blue and black Samsung TV box)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c4bf5605b8ee96c2bc8b80cc07da9c9ec71e1b5e8e5a904b4289cc9e8d3c23e"}
{"qa_id": "DAVIS__longboard__camera_translation_direction__00", "video_id": "DAVIS__longboard", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "up", "backward"], "frame_paths": ["media/29e03bb4abe24c28a35c/F0.jpg", "media/29e03bb4abe24c28a35c/F1.jpg", "media/29e03bb4abe24c28a35c/F2.jpg", "media/29e03bb4abe24c28a35c/F3.jpg", "media/29e03bb4abe24c28a35c/F4.jpg", "media/29e03bb4abe24c28a35c/F5.jpg", "media/29e03bb4abe24c28a35c/F6.jpg", "media/29e03bb4abe24c28a35c/F7.jpg", "media/29e03bb4abe24c28a35c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0071fa2d605111dbc9a11ec00f57534c836a31dab54af3a19dd71bcd11d8c6be"}
{"qa_id": "DAVIS__longboard__object_world_motion_direction__01", "video_id": "DAVIS__longboard", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in yellow shirt with black backpack) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "down", "backward", "forward"], "frame_paths": ["media/0b55d2a77118c04a6265/F0.jpg", "media/0b55d2a77118c04a6265/F1.jpg", "media/0b55d2a77118c04a6265/F2.jpg", "media/0b55d2a77118c04a6265/F3.jpg", "media/0b55d2a77118c04a6265/F4.jpg", "media/0b55d2a77118c04a6265/F5.jpg", "media/0b55d2a77118c04a6265/F6.jpg", "media/0b55d2a77118c04a6265/F7.jpg", "media/0b55d2a77118c04a6265/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in yellow shirt with black backpack) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f07b732e0a843deeba709386239b185dd1de85a4d1b7a5c85753e3b3d73cfa9c"}
{"qa_id": "DAVIS__motocross-bumps__camera_view_direction_change__00", "video_id": "DAVIS__motocross-bumps", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/712c42b0f23163fc90bd/F0.jpg", "media/712c42b0f23163fc90bd/F1.jpg", "media/712c42b0f23163fc90bd/F2.jpg", "media/712c42b0f23163fc90bd/F3.jpg", "media/712c42b0f23163fc90bd/F4.jpg", "media/712c42b0f23163fc90bd/F5.jpg", "media/712c42b0f23163fc90bd/F6.jpg", "media/712c42b0f23163fc90bd/F7.jpg", "media/712c42b0f23163fc90bd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88569cd745de284a24111d02b2fdec72688e7586531e7f8356c77faa9dd1fd28"}
{"qa_id": "DAVIS__motocross-bumps__object_world_motion_direction__01", "video_id": "DAVIS__motocross-bumps", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (motocross rider in black and white gear) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "backward", "up", "right"], "frame_paths": ["media/f56179ab55d6db857e18/F0.jpg", "media/f56179ab55d6db857e18/F1.jpg", "media/f56179ab55d6db857e18/F2.jpg", "media/f56179ab55d6db857e18/F3.jpg", "media/f56179ab55d6db857e18/F4.jpg", "media/f56179ab55d6db857e18/F5.jpg", "media/f56179ab55d6db857e18/F6.jpg", "media/f56179ab55d6db857e18/F7.jpg", "media/f56179ab55d6db857e18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (motocross rider in black and white gear) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a549d84cd18d383515a45b4d38cd662ddb056d4921990ad3e99480b4e9f45b0"}
{"qa_id": "DAVIS__motocross-bumps__camera_object_distance_change__02", "video_id": "DAVIS__motocross-bumps", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did object A (motocross rider in black and white gear) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/f56179ab55d6db857e18/F0.jpg", "media/f56179ab55d6db857e18/F1.jpg", "media/f56179ab55d6db857e18/F2.jpg", "media/f56179ab55d6db857e18/F3.jpg", "media/f56179ab55d6db857e18/F4.jpg", "media/f56179ab55d6db857e18/F5.jpg", "media/f56179ab55d6db857e18/F6.jpg", "media/f56179ab55d6db857e18/F7.jpg", "media/f56179ab55d6db857e18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (motocross rider in black and white gear) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68525b445b9a547e2e420f92051038ba16b4b126726a003643fa437c5a84c1a7"}
{"qa_id": "DAVIS__motocross-bumps__object_world_motion_direction__03", "video_id": "DAVIS__motocross-bumps", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object B (blue and white dirt bike) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "left", "forward", "up", "down"], "frame_paths": ["media/7b18ddfd9f5d6b87ede4/F0.jpg", "media/7b18ddfd9f5d6b87ede4/F1.jpg", "media/7b18ddfd9f5d6b87ede4/F2.jpg", "media/7b18ddfd9f5d6b87ede4/F3.jpg", "media/7b18ddfd9f5d6b87ede4/F4.jpg", "media/7b18ddfd9f5d6b87ede4/F5.jpg", "media/7b18ddfd9f5d6b87ede4/F6.jpg", "media/7b18ddfd9f5d6b87ede4/F7.jpg", "media/7b18ddfd9f5d6b87ede4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object B (blue and white dirt bike) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. left\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bbe6f6f1feca39657ba2048720c8be33a71edfe9e8c8222f89c849c2cc3a024e"}
{"qa_id": "DAVIS__motocross-bumps__camera_object_distance_change__04", "video_id": "DAVIS__motocross-bumps", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did object B (blue and white dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/7b18ddfd9f5d6b87ede4/F0.jpg", "media/7b18ddfd9f5d6b87ede4/F1.jpg", "media/7b18ddfd9f5d6b87ede4/F2.jpg", "media/7b18ddfd9f5d6b87ede4/F3.jpg", "media/7b18ddfd9f5d6b87ede4/F4.jpg", "media/7b18ddfd9f5d6b87ede4/F5.jpg", "media/7b18ddfd9f5d6b87ede4/F6.jpg", "media/7b18ddfd9f5d6b87ede4/F7.jpg", "media/7b18ddfd9f5d6b87ede4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object B (blue and white dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d148d37cc091e8a04d0878daac0df51b444c2f75431eed489467521abe1650e7"}
{"qa_id": "DAVIS__motorbike__camera_view_direction_change__00", "video_id": "DAVIS__motorbike", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/2ddf2324a61e695198cc/F0.jpg", "media/2ddf2324a61e695198cc/F1.jpg", "media/2ddf2324a61e695198cc/F2.jpg", "media/2ddf2324a61e695198cc/F3.jpg", "media/2ddf2324a61e695198cc/F4.jpg", "media/2ddf2324a61e695198cc/F5.jpg", "media/2ddf2324a61e695198cc/F6.jpg", "media/2ddf2324a61e695198cc/F7.jpg", "media/2ddf2324a61e695198cc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "DAVIS__motorbike__camera_object_distance_change__01", "video_id": "DAVIS__motorbike", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (motorcyclist in orange jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a3b4950ec3320490300c/F0.jpg", "media/a3b4950ec3320490300c/F1.jpg", "media/a3b4950ec3320490300c/F2.jpg", "media/a3b4950ec3320490300c/F3.jpg", "media/a3b4950ec3320490300c/F4.jpg", "media/a3b4950ec3320490300c/F5.jpg", "media/a3b4950ec3320490300c/F6.jpg", "media/a3b4950ec3320490300c/F7.jpg", "media/a3b4950ec3320490300c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (motorcyclist in orange jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "26ed8204abd96e85401ca38be3eba5f83af8fa6650e371c56a785c487c3a4583"}
{"qa_id": "DAVIS__motorbike__camera_object_distance_change__02", "video_id": "DAVIS__motorbike", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object B (motorcycle passenger in green jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/2cba5cf47d0166453db2/F0.jpg", "media/2cba5cf47d0166453db2/F1.jpg", "media/2cba5cf47d0166453db2/F2.jpg", "media/2cba5cf47d0166453db2/F3.jpg", "media/2cba5cf47d0166453db2/F4.jpg", "media/2cba5cf47d0166453db2/F5.jpg", "media/2cba5cf47d0166453db2/F6.jpg", "media/2cba5cf47d0166453db2/F7.jpg", "media/2cba5cf47d0166453db2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object B (motorcycle passenger in green jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc5c7bad456170809c14ce770f113c5d7d8596e293d8759b9405ea6ac2a0323b"}
{"qa_id": "DAVIS__night-race__camera_translation_direction__00", "video_id": "DAVIS__night-race", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "backward", "up", "forward"], "frame_paths": ["media/7efd262537b747cf1e18/F0.jpg", "media/7efd262537b747cf1e18/F1.jpg", "media/7efd262537b747cf1e18/F2.jpg", "media/7efd262537b747cf1e18/F3.jpg", "media/7efd262537b747cf1e18/F4.jpg", "media/7efd262537b747cf1e18/F5.jpg", "media/7efd262537b747cf1e18/F6.jpg", "media/7efd262537b747cf1e18/F7.jpg", "media/7efd262537b747cf1e18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87dd57b890c50d66a660ba0ae580c9d5337acc6a079b7963b1cfd74bae9a3dcd"}
{"qa_id": "DAVIS__night-race__object_world_motion_direction__01", "video_id": "DAVIS__night-race", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (dark sedan with orange outline) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "up", "forward", "backward"], "frame_paths": ["media/09d46f59c09d7ed5f527/F0.jpg", "media/09d46f59c09d7ed5f527/F1.jpg", "media/09d46f59c09d7ed5f527/F2.jpg", "media/09d46f59c09d7ed5f527/F3.jpg", "media/09d46f59c09d7ed5f527/F4.jpg", "media/09d46f59c09d7ed5f527/F5.jpg", "media/09d46f59c09d7ed5f527/F6.jpg", "media/09d46f59c09d7ed5f527/F7.jpg", "media/09d46f59c09d7ed5f527/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (dark sedan with orange outline) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3ff70fb606ac8b02547fd685b750376d7d8db15306f2c516220e382c017c9609"}
{"qa_id": "DAVIS__night-race__object_world_motion_direction__02", "video_id": "DAVIS__night-race", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (dark sedan with green outline) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "forward", "backward", "left", "right"], "frame_paths": ["media/dbd29f93aa6fb2b7985b/F0.jpg", "media/dbd29f93aa6fb2b7985b/F1.jpg", "media/dbd29f93aa6fb2b7985b/F2.jpg", "media/dbd29f93aa6fb2b7985b/F3.jpg", "media/dbd29f93aa6fb2b7985b/F4.jpg", "media/dbd29f93aa6fb2b7985b/F5.jpg", "media/dbd29f93aa6fb2b7985b/F6.jpg", "media/dbd29f93aa6fb2b7985b/F7.jpg", "media/dbd29f93aa6fb2b7985b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (dark sedan with green outline) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. forward\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fb76109d36a77bdd3afdd9136f08be93d357e80f67050f1dad11c94a02a0cd5"}
{"qa_id": "DAVIS__night-race__object_travel_distance_comparison__03", "video_id": "DAVIS__night-race", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark sedan with orange outline) or B (dark sedan with green outline)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/d5de48a2e87afbea4a47/F0.jpg", "media/d5de48a2e87afbea4a47/F1.jpg", "media/d5de48a2e87afbea4a47/F2.jpg", "media/d5de48a2e87afbea4a47/F3.jpg", "media/d5de48a2e87afbea4a47/F4.jpg", "media/d5de48a2e87afbea4a47/F5.jpg", "media/d5de48a2e87afbea4a47/F6.jpg", "media/d5de48a2e87afbea4a47/F7.jpg", "media/d5de48a2e87afbea4a47/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark sedan with orange outline) or B (dark sedan with green outline)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0080b2ab30e94ec568904fb5371e248a518b64b661b569f21c93d76123635900"}
{"qa_id": "DAVIS__paragliding-launch__camera_view_direction_change__00", "video_id": "DAVIS__paragliding-launch", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/0c9ee24835317584b0aa/F0.jpg", "media/0c9ee24835317584b0aa/F1.jpg", "media/0c9ee24835317584b0aa/F2.jpg", "media/0c9ee24835317584b0aa/F3.jpg", "media/0c9ee24835317584b0aa/F4.jpg", "media/0c9ee24835317584b0aa/F5.jpg", "media/0c9ee24835317584b0aa/F6.jpg", "media/0c9ee24835317584b0aa/F7.jpg", "media/0c9ee24835317584b0aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9557c0796e28ae2a4f32990ec2b747f1a5933b7979d70ae5cf47b924054e1483"}
{"qa_id": "DAVIS__paragliding-launch__object_world_motion_direction__01", "video_id": "DAVIS__paragliding-launch", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in dark flight suit and helmet paragliding) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "up", "down", "left"], "frame_paths": ["media/f4ac6bc936f53afbfa1e/F0.jpg", "media/f4ac6bc936f53afbfa1e/F1.jpg", "media/f4ac6bc936f53afbfa1e/F2.jpg", "media/f4ac6bc936f53afbfa1e/F3.jpg", "media/f4ac6bc936f53afbfa1e/F4.jpg", "media/f4ac6bc936f53afbfa1e/F5.jpg", "media/f4ac6bc936f53afbfa1e/F6.jpg", "media/f4ac6bc936f53afbfa1e/F7.jpg", "media/f4ac6bc936f53afbfa1e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in dark flight suit and helmet paragliding) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aaefdee7dbd5e1ef38017d045c5d257dac67afce98521bfcc1b340b84af99e76"}
{"qa_id": "DAVIS__scooter-black__camera_view_direction_change__00", "video_id": "DAVIS__scooter-black", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/2ff3decb388996f2bac4/F0.jpg", "media/2ff3decb388996f2bac4/F1.jpg", "media/2ff3decb388996f2bac4/F2.jpg", "media/2ff3decb388996f2bac4/F3.jpg", "media/2ff3decb388996f2bac4/F4.jpg", "media/2ff3decb388996f2bac4/F5.jpg", "media/2ff3decb388996f2bac4/F6.jpg", "media/2ff3decb388996f2bac4/F7.jpg", "media/2ff3decb388996f2bac4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f98011f6f5e3bef7856bffbee3a8c2579537b239284b48df76fd5bbadf4bc3f"}
{"qa_id": "DAVIS__scooter-black__camera_object_distance_change__01", "video_id": "DAVIS__scooter-black", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did object A (black scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/8e5c92d3fb8742b8b9b8/F0.jpg", "media/8e5c92d3fb8742b8b9b8/F1.jpg", "media/8e5c92d3fb8742b8b9b8/F2.jpg", "media/8e5c92d3fb8742b8b9b8/F3.jpg", "media/8e5c92d3fb8742b8b9b8/F4.jpg", "media/8e5c92d3fb8742b8b9b8/F5.jpg", "media/8e5c92d3fb8742b8b9b8/F6.jpg", "media/8e5c92d3fb8742b8b9b8/F7.jpg", "media/8e5c92d3fb8742b8b9b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (black scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "43be5ce8ea0bbc63cc0088f060ee1e06405c45ea9cf3ec2cd55ac3166f819aa0"}
{"qa_id": "DAVIS__scooter-black__camera_object_distance_change__02", "video_id": "DAVIS__scooter-black", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did object B (person wearing white helmet and dark jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7243c4ab9f8b625e00d7/F0.jpg", "media/7243c4ab9f8b625e00d7/F1.jpg", "media/7243c4ab9f8b625e00d7/F2.jpg", "media/7243c4ab9f8b625e00d7/F3.jpg", "media/7243c4ab9f8b625e00d7/F4.jpg", "media/7243c4ab9f8b625e00d7/F5.jpg", "media/7243c4ab9f8b625e00d7/F6.jpg", "media/7243c4ab9f8b625e00d7/F7.jpg", "media/7243c4ab9f8b625e00d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object B (person wearing white helmet and dark jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "60e88afefb09efb8d274029f069c57258c4a59bff9503cb63c3cc0d56e7babc5"}
{"qa_id": "DAVIS__scooter-gray__camera_view_direction_change__00", "video_id": "DAVIS__scooter-gray", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/c5d2b908a41a1e7ae8a3/F0.jpg", "media/c5d2b908a41a1e7ae8a3/F1.jpg", "media/c5d2b908a41a1e7ae8a3/F2.jpg", "media/c5d2b908a41a1e7ae8a3/F3.jpg", "media/c5d2b908a41a1e7ae8a3/F4.jpg", "media/c5d2b908a41a1e7ae8a3/F5.jpg", "media/c5d2b908a41a1e7ae8a3/F6.jpg", "media/c5d2b908a41a1e7ae8a3/F7.jpg", "media/c5d2b908a41a1e7ae8a3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db2dabb9c09d768aef279d407512cee59fa5581206d28df2cf3badd611191af1"}
{"qa_id": "DAVIS__skate-park__camera_view_direction_change__00", "video_id": "DAVIS__skate-park", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/c6e36381ab2bc031f0e5/F0.jpg", "media/c6e36381ab2bc031f0e5/F1.jpg", "media/c6e36381ab2bc031f0e5/F2.jpg", "media/c6e36381ab2bc031f0e5/F3.jpg", "media/c6e36381ab2bc031f0e5/F4.jpg", "media/c6e36381ab2bc031f0e5/F5.jpg", "media/c6e36381ab2bc031f0e5/F6.jpg", "media/c6e36381ab2bc031f0e5/F7.jpg", "media/c6e36381ab2bc031f0e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c70fea1841e5615b035b56b3517fe7bc5407eb2e748861736d937a6ed13d5c1"}
{"qa_id": "DAVIS__skate-park__object_world_motion_direction__01", "video_id": "DAVIS__skate-park", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person skateboarding in a concrete bowl) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "forward", "left", "right"], "frame_paths": ["media/ecaf038dd91185209753/F0.jpg", "media/ecaf038dd91185209753/F1.jpg", "media/ecaf038dd91185209753/F2.jpg", "media/ecaf038dd91185209753/F3.jpg", "media/ecaf038dd91185209753/F4.jpg", "media/ecaf038dd91185209753/F5.jpg", "media/ecaf038dd91185209753/F6.jpg", "media/ecaf038dd91185209753/F7.jpg", "media/ecaf038dd91185209753/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person skateboarding in a concrete bowl) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d163d426cabd1447c5dcd119d9d258fc54729ac1775a9fb10a8ddc7bfbd627b0"}
{"qa_id": "DAVIS__skate-park__camera_object_distance_change__02", "video_id": "DAVIS__skate-park", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F5, did object A (person skateboarding in a concrete bowl) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ecaf038dd91185209753/F0.jpg", "media/ecaf038dd91185209753/F1.jpg", "media/ecaf038dd91185209753/F2.jpg", "media/ecaf038dd91185209753/F3.jpg", "media/ecaf038dd91185209753/F4.jpg", "media/ecaf038dd91185209753/F5.jpg", "media/ecaf038dd91185209753/F6.jpg", "media/ecaf038dd91185209753/F7.jpg", "media/ecaf038dd91185209753/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (person skateboarding in a concrete bowl) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "899dd6ba999c919dc63cff4be1dc4212f52bf22a045590e8db973e3c36dff6f4"}
{"qa_id": "DAVIS__snowboard__object_world_motion_direction__00", "video_id": "DAVIS__snowboard", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (snowboarder in black jacket and helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "left", "down", "forward"], "frame_paths": ["media/f49a714f224724a13265/F0.jpg", "media/f49a714f224724a13265/F1.jpg", "media/f49a714f224724a13265/F2.jpg", "media/f49a714f224724a13265/F3.jpg", "media/f49a714f224724a13265/F4.jpg", "media/f49a714f224724a13265/F5.jpg", "media/f49a714f224724a13265/F6.jpg", "media/f49a714f224724a13265/F7.jpg", "media/f49a714f224724a13265/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (snowboarder in black jacket and helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44abaa7a5ba5fb50ae1e643f67926e852600d8ae2ec9fff4e1d8a090c12b14db"}
{"qa_id": "DAVIS__soapbox__camera_view_direction_change__00", "video_id": "DAVIS__soapbox", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/b8d7f8f9d8295ef0cf32/F0.jpg", "media/b8d7f8f9d8295ef0cf32/F1.jpg", "media/b8d7f8f9d8295ef0cf32/F2.jpg", "media/b8d7f8f9d8295ef0cf32/F3.jpg", "media/b8d7f8f9d8295ef0cf32/F4.jpg", "media/b8d7f8f9d8295ef0cf32/F5.jpg", "media/b8d7f8f9d8295ef0cf32/F6.jpg", "media/b8d7f8f9d8295ef0cf32/F7.jpg", "media/b8d7f8f9d8295ef0cf32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9557c0796e28ae2a4f32990ec2b747f1a5933b7979d70ae5cf47b924054e1483"}
{"qa_id": "DAVIS__soapbox__camera_object_distance_change__01", "video_id": "DAVIS__soapbox", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object B (person in green shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/0ce34df118f457b94ddf/F0.jpg", "media/0ce34df118f457b94ddf/F1.jpg", "media/0ce34df118f457b94ddf/F2.jpg", "media/0ce34df118f457b94ddf/F3.jpg", "media/0ce34df118f457b94ddf/F4.jpg", "media/0ce34df118f457b94ddf/F5.jpg", "media/0ce34df118f457b94ddf/F6.jpg", "media/0ce34df118f457b94ddf/F7.jpg", "media/0ce34df118f457b94ddf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object B (person in green shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "19aa6030d9f21e5c84ecd88d75588d0770354e8925f725964e51d63bf3acc9ca"}
{"qa_id": "DAVIS__soapbox__object_pair_distance_change__02", "video_id": "DAVIS__soapbox", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did the 3D distance between object A (person in orange shirt) and object B (person in green shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/64ea563b7cdd3bf6cb6c/F0.jpg", "media/64ea563b7cdd3bf6cb6c/F1.jpg", "media/64ea563b7cdd3bf6cb6c/F2.jpg", "media/64ea563b7cdd3bf6cb6c/F3.jpg", "media/64ea563b7cdd3bf6cb6c/F4.jpg", "media/64ea563b7cdd3bf6cb6c/F5.jpg", "media/64ea563b7cdd3bf6cb6c/F6.jpg", "media/64ea563b7cdd3bf6cb6c/F7.jpg", "media/64ea563b7cdd3bf6cb6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (person in orange shirt) and object B (person in green shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66491e246a428e3070b2aeff114c7c1a7728e2affaa54d2d1210000960a9f0e2"}
{"qa_id": "DAVIS__soapbox__object_travel_distance_comparison__03", "video_id": "DAVIS__soapbox", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt) or B (person in green shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/64ea563b7cdd3bf6cb6c/F0.jpg", "media/64ea563b7cdd3bf6cb6c/F1.jpg", "media/64ea563b7cdd3bf6cb6c/F2.jpg", "media/64ea563b7cdd3bf6cb6c/F3.jpg", "media/64ea563b7cdd3bf6cb6c/F4.jpg", "media/64ea563b7cdd3bf6cb6c/F5.jpg", "media/64ea563b7cdd3bf6cb6c/F6.jpg", "media/64ea563b7cdd3bf6cb6c/F7.jpg", "media/64ea563b7cdd3bf6cb6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt) or B (person in green shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "308e198ce2dc87211997aabf894b785749470553baa48057e66e74a76e0167a5"}
{"qa_id": "DAVIS__stroller__camera_view_direction_change__00", "video_id": "DAVIS__stroller", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/fb020aa407b501de9922/F0.jpg", "media/fb020aa407b501de9922/F1.jpg", "media/fb020aa407b501de9922/F2.jpg", "media/fb020aa407b501de9922/F3.jpg", "media/fb020aa407b501de9922/F4.jpg", "media/fb020aa407b501de9922/F5.jpg", "media/fb020aa407b501de9922/F6.jpg", "media/fb020aa407b501de9922/F7.jpg", "media/fb020aa407b501de9922/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "DAVIS__stroller__camera_object_distance_change__01", "video_id": "DAVIS__stroller", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object A (blue stroller with a dark canopy) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a5c74e263c55eac74d94/F0.jpg", "media/a5c74e263c55eac74d94/F1.jpg", "media/a5c74e263c55eac74d94/F2.jpg", "media/a5c74e263c55eac74d94/F3.jpg", "media/a5c74e263c55eac74d94/F4.jpg", "media/a5c74e263c55eac74d94/F5.jpg", "media/a5c74e263c55eac74d94/F6.jpg", "media/a5c74e263c55eac74d94/F7.jpg", "media/a5c74e263c55eac74d94/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (blue stroller with a dark canopy) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "61eb88fdc5e63ddfe6a5685ca99505912724c8fc95e265cd72245d544ad5d264"}
{"qa_id": "DAVIS__stroller__camera_object_distance_change__02", "video_id": "DAVIS__stroller", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object B (toddler sitting in the stroller) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/92146c8f38a4edc73c77/F0.jpg", "media/92146c8f38a4edc73c77/F1.jpg", "media/92146c8f38a4edc73c77/F2.jpg", "media/92146c8f38a4edc73c77/F3.jpg", "media/92146c8f38a4edc73c77/F4.jpg", "media/92146c8f38a4edc73c77/F5.jpg", "media/92146c8f38a4edc73c77/F6.jpg", "media/92146c8f38a4edc73c77/F7.jpg", "media/92146c8f38a4edc73c77/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (toddler sitting in the stroller) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5e817f26138e70311c253efcdaf314acd9ceb430a0045deccb2647d895fed4f"}
{"qa_id": "DAVIS__stunt__camera_view_direction_change__00", "video_id": "DAVIS__stunt", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/e81c37764a38c5f51400/F0.jpg", "media/e81c37764a38c5f51400/F1.jpg", "media/e81c37764a38c5f51400/F2.jpg", "media/e81c37764a38c5f51400/F3.jpg", "media/e81c37764a38c5f51400/F4.jpg", "media/e81c37764a38c5f51400/F5.jpg", "media/e81c37764a38c5f51400/F6.jpg", "media/e81c37764a38c5f51400/F7.jpg", "media/e81c37764a38c5f51400/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a245e62b618f2ca15b6ff17fcece0172f5241f03da0191cbc429ffde38e21fc8"}
{"qa_id": "DAVIS__stunt__camera_object_distance_change__01", "video_id": "DAVIS__stunt", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (person in blue shirt and black pants riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/19c5148725b03bc170ec/F0.jpg", "media/19c5148725b03bc170ec/F1.jpg", "media/19c5148725b03bc170ec/F2.jpg", "media/19c5148725b03bc170ec/F3.jpg", "media/19c5148725b03bc170ec/F4.jpg", "media/19c5148725b03bc170ec/F5.jpg", "media/19c5148725b03bc170ec/F6.jpg", "media/19c5148725b03bc170ec/F7.jpg", "media/19c5148725b03bc170ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in blue shirt and black pants riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f2452336c3826c4f1fd567de7a7d891b530fa77a01805168d1db43c4fd8b3ca"}
{"qa_id": "DAVIS__swing__object_world_motion_direction__00", "video_id": "DAVIS__swing", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in white shirt and blue jeans sitting on a swing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/459fbe438b9c8dc11e15/F0.jpg", "media/459fbe438b9c8dc11e15/F1.jpg", "media/459fbe438b9c8dc11e15/F2.jpg", "media/459fbe438b9c8dc11e15/F3.jpg", "media/459fbe438b9c8dc11e15/F4.jpg", "media/459fbe438b9c8dc11e15/F5.jpg", "media/459fbe438b9c8dc11e15/F6.jpg", "media/459fbe438b9c8dc11e15/F7.jpg", "media/459fbe438b9c8dc11e15/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in white shirt and blue jeans sitting on a swing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68c4f79c4834f4b844929c5ca7e18e9b1d69ea933d66598f6f6c2b07f4e69de3"}
{"qa_id": "DAVIS__tractor-sand__object_world_motion_direction__00", "video_id": "DAVIS__tractor-sand", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object A (person riding a small tracked vehicle) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "forward", "down"], "frame_paths": ["media/968a7a169b6235eb0296/F0.jpg", "media/968a7a169b6235eb0296/F1.jpg", "media/968a7a169b6235eb0296/F2.jpg", "media/968a7a169b6235eb0296/F3.jpg", "media/968a7a169b6235eb0296/F4.jpg", "media/968a7a169b6235eb0296/F5.jpg", "media/968a7a169b6235eb0296/F6.jpg", "media/968a7a169b6235eb0296/F7.jpg", "media/968a7a169b6235eb0296/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object A (person riding a small tracked vehicle) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "975d5c5c572ce5924a1cc78a667d1f8aef5abb2cb0062be993b4abbd0067c2f2"}
{"qa_id": "DAVIS__tractor-sand__camera_object_distance_change__01", "video_id": "DAVIS__tractor-sand", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (person riding a small tracked vehicle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/968a7a169b6235eb0296/F0.jpg", "media/968a7a169b6235eb0296/F1.jpg", "media/968a7a169b6235eb0296/F2.jpg", "media/968a7a169b6235eb0296/F3.jpg", "media/968a7a169b6235eb0296/F4.jpg", "media/968a7a169b6235eb0296/F5.jpg", "media/968a7a169b6235eb0296/F6.jpg", "media/968a7a169b6235eb0296/F7.jpg", "media/968a7a169b6235eb0296/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (person riding a small tracked vehicle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc6d23e6c81e67d59fdd47968e8ca7ac62a97d12d2a11c1b840d2edb18d2c490"}
{"qa_id": "DAVIS__train__camera_translation_direction__00", "video_id": "DAVIS__train", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "forward", "down", "up"], "frame_paths": ["media/0c61837c8bf1cb05e719/F0.jpg", "media/0c61837c8bf1cb05e719/F1.jpg", "media/0c61837c8bf1cb05e719/F2.jpg", "media/0c61837c8bf1cb05e719/F3.jpg", "media/0c61837c8bf1cb05e719/F4.jpg", "media/0c61837c8bf1cb05e719/F5.jpg", "media/0c61837c8bf1cb05e719/F6.jpg", "media/0c61837c8bf1cb05e719/F7.jpg", "media/0c61837c8bf1cb05e719/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "32968ab4e177e5a6de717b4d31e37c680451bd1913804e76faf013e840480ffb"}
{"qa_id": "DAVIS__train__object_world_motion_direction__01", "video_id": "DAVIS__train", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (grey toy train car with a flat roof) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/105e48adcd23f3be2971/F0.jpg", "media/105e48adcd23f3be2971/F1.jpg", "media/105e48adcd23f3be2971/F2.jpg", "media/105e48adcd23f3be2971/F3.jpg", "media/105e48adcd23f3be2971/F4.jpg", "media/105e48adcd23f3be2971/F5.jpg", "media/105e48adcd23f3be2971/F6.jpg", "media/105e48adcd23f3be2971/F7.jpg", "media/105e48adcd23f3be2971/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (grey toy train car with a flat roof) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d3dbe8db54ccde58071607235a8f7d369d997672a30c04babe28627d75166d9"}
{"qa_id": "DAVIS__train__object_world_motion_direction__02", "video_id": "DAVIS__train", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (blue toy train engine) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "down", "forward", "up"], "frame_paths": ["media/9e0fe48d0d5b59b7b460/F0.jpg", "media/9e0fe48d0d5b59b7b460/F1.jpg", "media/9e0fe48d0d5b59b7b460/F2.jpg", "media/9e0fe48d0d5b59b7b460/F3.jpg", "media/9e0fe48d0d5b59b7b460/F4.jpg", "media/9e0fe48d0d5b59b7b460/F5.jpg", "media/9e0fe48d0d5b59b7b460/F6.jpg", "media/9e0fe48d0d5b59b7b460/F7.jpg", "media/9e0fe48d0d5b59b7b460/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (blue toy train engine) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. down\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "807dced2fb36380854cac19fc63866b476625831a5935caefa4781d072d6f8e4"}
{"qa_id": "DAVIS__train__camera_object_distance_change__03", "video_id": "DAVIS__train", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object B (blue toy train engine) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9e0fe48d0d5b59b7b460/F0.jpg", "media/9e0fe48d0d5b59b7b460/F1.jpg", "media/9e0fe48d0d5b59b7b460/F2.jpg", "media/9e0fe48d0d5b59b7b460/F3.jpg", "media/9e0fe48d0d5b59b7b460/F4.jpg", "media/9e0fe48d0d5b59b7b460/F5.jpg", "media/9e0fe48d0d5b59b7b460/F6.jpg", "media/9e0fe48d0d5b59b7b460/F7.jpg", "media/9e0fe48d0d5b59b7b460/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (blue toy train engine) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d93f2e870d46501f0400a1a2b68cfed966670b0a2288dd1396fb44e53da9c394"}
{"qa_id": "DAVIS__walking__camera_translation_direction__00", "video_id": "DAVIS__walking", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "right", "up", "down", "left"], "frame_paths": ["media/35f6734d4d821032176f/F0.jpg", "media/35f6734d4d821032176f/F1.jpg", "media/35f6734d4d821032176f/F2.jpg", "media/35f6734d4d821032176f/F3.jpg", "media/35f6734d4d821032176f/F4.jpg", "media/35f6734d4d821032176f/F5.jpg", "media/35f6734d4d821032176f/F6.jpg", "media/35f6734d4d821032176f/F7.jpg", "media/35f6734d4d821032176f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5eb18b0394e453aa95a020cb890593d5303ee68aea9775de7d5d33ac3e26c99"}
{"qa_id": "DAVIS__walking__camera_object_distance_change__01", "video_id": "DAVIS__walking", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (man with a beard wearing a blue blazer and black shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/293b45edfdfa20059972/F0.jpg", "media/293b45edfdfa20059972/F1.jpg", "media/293b45edfdfa20059972/F2.jpg", "media/293b45edfdfa20059972/F3.jpg", "media/293b45edfdfa20059972/F4.jpg", "media/293b45edfdfa20059972/F5.jpg", "media/293b45edfdfa20059972/F6.jpg", "media/293b45edfdfa20059972/F7.jpg", "media/293b45edfdfa20059972/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (man with a beard wearing a blue blazer and black shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ae098a1a6d71cd9f05c063db7dcd797bfaf17d2f1f2086955614d80cfde244c6"}
{"qa_id": "MOSE__001ca3cb__camera_object_distance_change__00", "video_id": "MOSE__001ca3cb", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object A (pink fish) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/108bef36f4b50c64b12a/F0.jpg", "media/108bef36f4b50c64b12a/F1.jpg", "media/108bef36f4b50c64b12a/F2.jpg", "media/108bef36f4b50c64b12a/F3.jpg", "media/108bef36f4b50c64b12a/F4.jpg", "media/108bef36f4b50c64b12a/F5.jpg", "media/108bef36f4b50c64b12a/F6.jpg", "media/108bef36f4b50c64b12a/F7.jpg", "media/108bef36f4b50c64b12a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (pink fish) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2f0cd34c3fc986cb07c74b612925c361b05e6646fd4a9f5980dd328c6413a145"}
{"qa_id": "MOSE__00501424__object_world_motion_direction__00", "video_id": "MOSE__00501424", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (dark brown horse with a small white marking on its forehead) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "backward", "down", "left"], "frame_paths": ["media/a7f3b40f80026e8f66cd/F0.jpg", "media/a7f3b40f80026e8f66cd/F1.jpg", "media/a7f3b40f80026e8f66cd/F2.jpg", "media/a7f3b40f80026e8f66cd/F3.jpg", "media/a7f3b40f80026e8f66cd/F4.jpg", "media/a7f3b40f80026e8f66cd/F5.jpg", "media/a7f3b40f80026e8f66cd/F6.jpg", "media/a7f3b40f80026e8f66cd/F7.jpg", "media/a7f3b40f80026e8f66cd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (dark brown horse with a small white marking on its forehead) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f185140b123d831187a9032effa9e3d653739e76750fd13daa777541d393bd94"}
{"qa_id": "MOSE__00501424__camera_object_distance_change__01", "video_id": "MOSE__00501424", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (dark brown horse with a small white marking on its forehead) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a7f3b40f80026e8f66cd/F0.jpg", "media/a7f3b40f80026e8f66cd/F1.jpg", "media/a7f3b40f80026e8f66cd/F2.jpg", "media/a7f3b40f80026e8f66cd/F3.jpg", "media/a7f3b40f80026e8f66cd/F4.jpg", "media/a7f3b40f80026e8f66cd/F5.jpg", "media/a7f3b40f80026e8f66cd/F6.jpg", "media/a7f3b40f80026e8f66cd/F7.jpg", "media/a7f3b40f80026e8f66cd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (dark brown horse with a small white marking on its forehead) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "021a1a438790bc40bee84804b9a4d77f0d9d915856188b3eb8c50f11851026ec"}
{"qa_id": "MOSE__00501424__object_world_motion_direction__02", "video_id": "MOSE__00501424", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (dark brown horse on the far left of the group) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "forward", "up", "down", "right"], "frame_paths": ["media/601e1f3f547be5ad7dfc/F0.jpg", "media/601e1f3f547be5ad7dfc/F1.jpg", "media/601e1f3f547be5ad7dfc/F2.jpg", "media/601e1f3f547be5ad7dfc/F3.jpg", "media/601e1f3f547be5ad7dfc/F4.jpg", "media/601e1f3f547be5ad7dfc/F5.jpg", "media/601e1f3f547be5ad7dfc/F6.jpg", "media/601e1f3f547be5ad7dfc/F7.jpg", "media/601e1f3f547be5ad7dfc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (dark brown horse on the far left of the group) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. forward\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5e668da5afeed0974f90cfa30c4bcc861ac2013debc743a2391c952b544d20a"}
{"qa_id": "MOSE__00501424__camera_object_distance_change__03", "video_id": "MOSE__00501424", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object B (dark brown horse on the far left of the group) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/601e1f3f547be5ad7dfc/F0.jpg", "media/601e1f3f547be5ad7dfc/F1.jpg", "media/601e1f3f547be5ad7dfc/F2.jpg", "media/601e1f3f547be5ad7dfc/F3.jpg", "media/601e1f3f547be5ad7dfc/F4.jpg", "media/601e1f3f547be5ad7dfc/F5.jpg", "media/601e1f3f547be5ad7dfc/F6.jpg", "media/601e1f3f547be5ad7dfc/F7.jpg", "media/601e1f3f547be5ad7dfc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (dark brown horse on the far left of the group) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e47eb06ba5d4702bfbdbd19166890a3fc8dcb5bf8a3e33743467c310a4fdf5a"}
{"qa_id": "MOSE__02221fb0__object_pair_distance_change__00", "video_id": "MOSE__02221fb0", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (silver sedan) and object B (person in dark clothing) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f272aac250ed07deb3d1/F0.jpg", "media/f272aac250ed07deb3d1/F1.jpg", "media/f272aac250ed07deb3d1/F2.jpg", "media/f272aac250ed07deb3d1/F3.jpg", "media/f272aac250ed07deb3d1/F4.jpg", "media/f272aac250ed07deb3d1/F5.jpg", "media/f272aac250ed07deb3d1/F6.jpg", "media/f272aac250ed07deb3d1/F7.jpg", "media/f272aac250ed07deb3d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (silver sedan) and object B (person in dark clothing) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d56354e7ba89f5243a836a753b433a6f56df34e32c9ee8617774e6d1b85006ef"}
{"qa_id": "MOSE__02221fb0__object_travel_distance_comparison__01", "video_id": "MOSE__02221fb0", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (silver sedan) or B (person in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/f272aac250ed07deb3d1/F0.jpg", "media/f272aac250ed07deb3d1/F1.jpg", "media/f272aac250ed07deb3d1/F2.jpg", "media/f272aac250ed07deb3d1/F3.jpg", "media/f272aac250ed07deb3d1/F4.jpg", "media/f272aac250ed07deb3d1/F5.jpg", "media/f272aac250ed07deb3d1/F6.jpg", "media/f272aac250ed07deb3d1/F7.jpg", "media/f272aac250ed07deb3d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (silver sedan) or B (person in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c4b47e07df94b6b3fe22a09645e298bb78316b653293fd0ab9120e4919131775"}
{"qa_id": "MOSE__028f6f64__camera_translation_direction__00", "video_id": "MOSE__028f6f64", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "right", "left"], "frame_paths": ["media/43db61a5e1771b33f006/F0.jpg", "media/43db61a5e1771b33f006/F1.jpg", "media/43db61a5e1771b33f006/F2.jpg", "media/43db61a5e1771b33f006/F3.jpg", "media/43db61a5e1771b33f006/F4.jpg", "media/43db61a5e1771b33f006/F5.jpg", "media/43db61a5e1771b33f006/F6.jpg", "media/43db61a5e1771b33f006/F7.jpg", "media/43db61a5e1771b33f006/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c157d1bcf69fcd0779e18cee8d7faa64161a0f4bb313b6ad630c4a20c0dafe3"}
{"qa_id": "MOSE__02deca50__camera_view_direction_change__00", "video_id": "MOSE__02deca50", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/1a3191ca221b06e3c2aa/F0.jpg", "media/1a3191ca221b06e3c2aa/F1.jpg", "media/1a3191ca221b06e3c2aa/F2.jpg", "media/1a3191ca221b06e3c2aa/F3.jpg", "media/1a3191ca221b06e3c2aa/F4.jpg", "media/1a3191ca221b06e3c2aa/F5.jpg", "media/1a3191ca221b06e3c2aa/F6.jpg", "media/1a3191ca221b06e3c2aa/F7.jpg", "media/1a3191ca221b06e3c2aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "MOSE__02deca50__object_world_motion_direction__01", "video_id": "MOSE__02deca50", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (zebra on the right with head lowered) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "up", "forward", "backward"], "frame_paths": ["media/1058aecfb0b3d1fe6507/F0.jpg", "media/1058aecfb0b3d1fe6507/F1.jpg", "media/1058aecfb0b3d1fe6507/F2.jpg", "media/1058aecfb0b3d1fe6507/F3.jpg", "media/1058aecfb0b3d1fe6507/F4.jpg", "media/1058aecfb0b3d1fe6507/F5.jpg", "media/1058aecfb0b3d1fe6507/F6.jpg", "media/1058aecfb0b3d1fe6507/F7.jpg", "media/1058aecfb0b3d1fe6507/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (zebra on the right with head lowered) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e43e8fbde2e5151e103c87e218ff54a0779ca62fd31a557f27781526c4a599ac"}
{"qa_id": "MOSE__02deca50__object_world_motion_direction__02", "video_id": "MOSE__02deca50", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (zebra on the left with head raised) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "forward", "left", "backward"], "frame_paths": ["media/e31e5a952f8432588f6a/F0.jpg", "media/e31e5a952f8432588f6a/F1.jpg", "media/e31e5a952f8432588f6a/F2.jpg", "media/e31e5a952f8432588f6a/F3.jpg", "media/e31e5a952f8432588f6a/F4.jpg", "media/e31e5a952f8432588f6a/F5.jpg", "media/e31e5a952f8432588f6a/F6.jpg", "media/e31e5a952f8432588f6a/F7.jpg", "media/e31e5a952f8432588f6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (zebra on the left with head raised) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. forward\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bd12fb4afe5d41b700a6164c9b9dae6a9d74c144ed69a9c5bb051574a07aaca2"}
{"qa_id": "MOSE__033333fd__object_world_motion_direction__01", "video_id": "MOSE__033333fd", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (smaller juvenile elephant walking alongside the adult) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "up", "left", "backward"], "frame_paths": ["media/84857efa1c4e5f329650/F0.jpg", "media/84857efa1c4e5f329650/F1.jpg", "media/84857efa1c4e5f329650/F2.jpg", "media/84857efa1c4e5f329650/F3.jpg", "media/84857efa1c4e5f329650/F4.jpg", "media/84857efa1c4e5f329650/F5.jpg", "media/84857efa1c4e5f329650/F6.jpg", "media/84857efa1c4e5f329650/F7.jpg", "media/84857efa1c4e5f329650/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (smaller juvenile elephant walking alongside the adult) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. up\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "069a254803a0fe4b25c589717527715d1850d14efb469a1bae91c9af61cb3e6a"}
{"qa_id": "MOSE__03c99e83__camera_view_direction_change__00", "video_id": "MOSE__03c99e83", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/db300d507e3db795fc6d/F0.jpg", "media/db300d507e3db795fc6d/F1.jpg", "media/db300d507e3db795fc6d/F2.jpg", "media/db300d507e3db795fc6d/F3.jpg", "media/db300d507e3db795fc6d/F4.jpg", "media/db300d507e3db795fc6d/F5.jpg", "media/db300d507e3db795fc6d/F6.jpg", "media/db300d507e3db795fc6d/F7.jpg", "media/db300d507e3db795fc6d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48b0dbe5136add4561fdd3af171b3ed657216b06d587528f3932aa08151751c1"}
{"qa_id": "MOSE__03e0f9b8__camera_translation_direction__00", "video_id": "MOSE__03e0f9b8", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/1bf65a4b5950e672ab5e/F0.jpg", "media/1bf65a4b5950e672ab5e/F1.jpg", "media/1bf65a4b5950e672ab5e/F2.jpg", "media/1bf65a4b5950e672ab5e/F3.jpg", "media/1bf65a4b5950e672ab5e/F4.jpg", "media/1bf65a4b5950e672ab5e/F5.jpg", "media/1bf65a4b5950e672ab5e/F6.jpg", "media/1bf65a4b5950e672ab5e/F7.jpg", "media/1bf65a4b5950e672ab5e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c705eb8808b6926e3af64c83d568f8e78de279eb30c39ec22b2aa303d56db961"}
{"qa_id": "MOSE__03e0f9b8__object_pair_distance_change__01", "video_id": "MOSE__03e0f9b8", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (orange fish with a prominent head bump) and object B (small yellowish fish near the gravel) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/dafb6c9d87a58aae3baa/F0.jpg", "media/dafb6c9d87a58aae3baa/F1.jpg", "media/dafb6c9d87a58aae3baa/F2.jpg", "media/dafb6c9d87a58aae3baa/F3.jpg", "media/dafb6c9d87a58aae3baa/F4.jpg", "media/dafb6c9d87a58aae3baa/F5.jpg", "media/dafb6c9d87a58aae3baa/F6.jpg", "media/dafb6c9d87a58aae3baa/F7.jpg", "media/dafb6c9d87a58aae3baa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (orange fish with a prominent head bump) and object B (small yellowish fish near the gravel) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e131a074b7e12c57b16cbf57f21edc6af98851ecd963e7f3cdc8e9f3f58e7800"}
{"qa_id": "MOSE__03e5f069__camera_translation_direction__00", "video_id": "MOSE__03e5f069", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/8c309e30bbf166d7958f/F0.jpg", "media/8c309e30bbf166d7958f/F1.jpg", "media/8c309e30bbf166d7958f/F2.jpg", "media/8c309e30bbf166d7958f/F3.jpg", "media/8c309e30bbf166d7958f/F4.jpg", "media/8c309e30bbf166d7958f/F5.jpg", "media/8c309e30bbf166d7958f/F6.jpg", "media/8c309e30bbf166d7958f/F7.jpg", "media/8c309e30bbf166d7958f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d599dd2b4f48439d8e713e1e7bc9f237edcf88e68f9cb3382850ce0c65a097a0"}
{"qa_id": "MOSE__03e5f069__camera_object_distance_change__01", "video_id": "MOSE__03e5f069", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object B (person in white t-shirt and light shorts riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/59d270e09a79c07f0209/F0.jpg", "media/59d270e09a79c07f0209/F1.jpg", "media/59d270e09a79c07f0209/F2.jpg", "media/59d270e09a79c07f0209/F3.jpg", "media/59d270e09a79c07f0209/F4.jpg", "media/59d270e09a79c07f0209/F5.jpg", "media/59d270e09a79c07f0209/F6.jpg", "media/59d270e09a79c07f0209/F7.jpg", "media/59d270e09a79c07f0209/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object B (person in white t-shirt and light shorts riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8f6cae6e05cf8658cf393c6c32e965762d4ea700c4929bcc160ce51e2bdf58a"}
{"qa_id": "MOSE__03e5f069__object_pair_distance_change__02", "video_id": "MOSE__03e5f069", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (person in white shirt riding a bicycle) and object B (person in white t-shirt and light shorts riding a bicycle) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/45423fb71e5c8599af60/F0.jpg", "media/45423fb71e5c8599af60/F1.jpg", "media/45423fb71e5c8599af60/F2.jpg", "media/45423fb71e5c8599af60/F3.jpg", "media/45423fb71e5c8599af60/F4.jpg", "media/45423fb71e5c8599af60/F5.jpg", "media/45423fb71e5c8599af60/F6.jpg", "media/45423fb71e5c8599af60/F7.jpg", "media/45423fb71e5c8599af60/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (person in white shirt riding a bicycle) and object B (person in white t-shirt and light shorts riding a bicycle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "baa97e31aafa177904e53dc12723d127ee317dddbf816a857444358eeb6c5c4b"}
{"qa_id": "MOSE__0442a954__object_world_motion_direction__00", "video_id": "MOSE__0442a954", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in white shirt and orange shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "right", "forward"], "frame_paths": ["media/922117d8df57afeeae8f/F0.jpg", "media/922117d8df57afeeae8f/F1.jpg", "media/922117d8df57afeeae8f/F2.jpg", "media/922117d8df57afeeae8f/F3.jpg", "media/922117d8df57afeeae8f/F4.jpg", "media/922117d8df57afeeae8f/F5.jpg", "media/922117d8df57afeeae8f/F6.jpg", "media/922117d8df57afeeae8f/F7.jpg", "media/922117d8df57afeeae8f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g7"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in white shirt and orange shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f29e8e552e317c709959669e32dccc3b547fbff5aaea0835bcc3d2e8ecb9cfa"}
{"qa_id": "MOSE__0442a954__object_pair_distance_change__01", "video_id": "MOSE__0442a954", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did the 3D distance between object A (person in white shirt and orange shorts) and object B (person in red shirt with number 11) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/92ca0956d9a1f73d2417/F0.jpg", "media/92ca0956d9a1f73d2417/F1.jpg", "media/92ca0956d9a1f73d2417/F2.jpg", "media/92ca0956d9a1f73d2417/F3.jpg", "media/92ca0956d9a1f73d2417/F4.jpg", "media/92ca0956d9a1f73d2417/F5.jpg", "media/92ca0956d9a1f73d2417/F6.jpg", "media/92ca0956d9a1f73d2417/F7.jpg", "media/92ca0956d9a1f73d2417/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g7"}, "requested_object_ids": ["g4", "g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (person in white shirt and orange shorts) and object B (person in red shirt with number 11) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "79bec81d5c16cd8ec6ef454f1208d1c9dfec40f82141eaab189898ebedd7ab20"}
{"qa_id": "MOSE__04ae7a6b__camera_translation_direction__00", "video_id": "MOSE__04ae7a6b", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/c0fe2637d558066ccaa3/F0.jpg", "media/c0fe2637d558066ccaa3/F1.jpg", "media/c0fe2637d558066ccaa3/F2.jpg", "media/c0fe2637d558066ccaa3/F3.jpg", "media/c0fe2637d558066ccaa3/F4.jpg", "media/c0fe2637d558066ccaa3/F5.jpg", "media/c0fe2637d558066ccaa3/F6.jpg", "media/c0fe2637d558066ccaa3/F7.jpg", "media/c0fe2637d558066ccaa3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5cd3f8706a92139f6f8505332a5cc4a4f13b67492be677dc9e6aadb17de7d2d"}
{"qa_id": "MOSE__04ae7a6b__camera_view_direction_change__01", "video_id": "MOSE__04ae7a6b", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/c0fe2637d558066ccaa3/F0.jpg", "media/c0fe2637d558066ccaa3/F1.jpg", "media/c0fe2637d558066ccaa3/F2.jpg", "media/c0fe2637d558066ccaa3/F3.jpg", "media/c0fe2637d558066ccaa3/F4.jpg", "media/c0fe2637d558066ccaa3/F5.jpg", "media/c0fe2637d558066ccaa3/F6.jpg", "media/c0fe2637d558066ccaa3/F7.jpg", "media/c0fe2637d558066ccaa3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "MOSE__051052d8__camera_view_direction_change__00", "video_id": "MOSE__051052d8", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/226282d44fee14ed3d12/F0.jpg", "media/226282d44fee14ed3d12/F1.jpg", "media/226282d44fee14ed3d12/F2.jpg", "media/226282d44fee14ed3d12/F3.jpg", "media/226282d44fee14ed3d12/F4.jpg", "media/226282d44fee14ed3d12/F5.jpg", "media/226282d44fee14ed3d12/F6.jpg", "media/226282d44fee14ed3d12/F7.jpg", "media/226282d44fee14ed3d12/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16a4ca6eeb0d84a0d96a1b095761436c4777b97628640bbab69e354a88eb5d55"}
{"qa_id": "MOSE__051052d8__object_world_motion_direction__01", "video_id": "MOSE__051052d8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (large elephant with head lowered in the grass) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/4e3484a7721dd0b4f153/F0.jpg", "media/4e3484a7721dd0b4f153/F1.jpg", "media/4e3484a7721dd0b4f153/F2.jpg", "media/4e3484a7721dd0b4f153/F3.jpg", "media/4e3484a7721dd0b4f153/F4.jpg", "media/4e3484a7721dd0b4f153/F5.jpg", "media/4e3484a7721dd0b4f153/F6.jpg", "media/4e3484a7721dd0b4f153/F7.jpg", "media/4e3484a7721dd0b4f153/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (large elephant with head lowered in the grass) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "018dc509c22161050474d41114552a97d146a16eedb001dba40cf49b0a08d163"}
{"qa_id": "MOSE__051052d8__object_travel_distance_comparison__02", "video_id": "MOSE__051052d8", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant with head lowered in the grass) or B (large elephant standing behind g1)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/25956838c01f85530a50/F0.jpg", "media/25956838c01f85530a50/F1.jpg", "media/25956838c01f85530a50/F2.jpg", "media/25956838c01f85530a50/F3.jpg", "media/25956838c01f85530a50/F4.jpg", "media/25956838c01f85530a50/F5.jpg", "media/25956838c01f85530a50/F6.jpg", "media/25956838c01f85530a50/F7.jpg", "media/25956838c01f85530a50/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant with head lowered in the grass) or B (large elephant standing behind g1)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "baef78cbf9c53e26ce3e920802327d5f3364a3540ebe33480dbd2e42fd8f2677"}
{"qa_id": "MOSE__05ed1023__object_pair_distance_change__00", "video_id": "MOSE__05ed1023", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (tiger drinking water at the edge of a stream) and object B (tiger walking on the rocky bank in the background) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/7c862eab94c7ea78d927/F0.jpg", "media/7c862eab94c7ea78d927/F1.jpg", "media/7c862eab94c7ea78d927/F2.jpg", "media/7c862eab94c7ea78d927/F3.jpg", "media/7c862eab94c7ea78d927/F4.jpg", "media/7c862eab94c7ea78d927/F5.jpg", "media/7c862eab94c7ea78d927/F6.jpg", "media/7c862eab94c7ea78d927/F7.jpg", "media/7c862eab94c7ea78d927/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (tiger drinking water at the edge of a stream) and object B (tiger walking on the rocky bank in the background) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "babd2ad69e0b9abb5946bb310abc6521f13debee7266bc4ed4773ebdc56a807a"}
{"qa_id": "MOSE__06273084__camera_view_direction_change__00", "video_id": "MOSE__06273084", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/68238a3d488bbe15584c/F0.jpg", "media/68238a3d488bbe15584c/F1.jpg", "media/68238a3d488bbe15584c/F2.jpg", "media/68238a3d488bbe15584c/F3.jpg", "media/68238a3d488bbe15584c/F4.jpg", "media/68238a3d488bbe15584c/F5.jpg", "media/68238a3d488bbe15584c/F6.jpg", "media/68238a3d488bbe15584c/F7.jpg", "media/68238a3d488bbe15584c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "648736a9959518de0e09fa475f2a5bbc01a36b254b1af7e9c406fc2e9bb070a1"}
{"qa_id": "MOSE__06273084__object_world_motion_direction__01", "video_id": "MOSE__06273084", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (brown horse running) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "left", "up", "down", "right"], "frame_paths": ["media/8f8395c4b4f53d604df8/F0.jpg", "media/8f8395c4b4f53d604df8/F1.jpg", "media/8f8395c4b4f53d604df8/F2.jpg", "media/8f8395c4b4f53d604df8/F3.jpg", "media/8f8395c4b4f53d604df8/F4.jpg", "media/8f8395c4b4f53d604df8/F5.jpg", "media/8f8395c4b4f53d604df8/F6.jpg", "media/8f8395c4b4f53d604df8/F7.jpg", "media/8f8395c4b4f53d604df8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (brown horse running) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. left\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "814a7c40a2623828469dbd84696a5d3ced9e8feccb5fd673e923aa84bc14dacd"}
{"qa_id": "MOSE__072e7b3f__camera_translation_direction__00", "video_id": "MOSE__072e7b3f", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/caeba79eb00861d0b3f4/F0.jpg", "media/caeba79eb00861d0b3f4/F1.jpg", "media/caeba79eb00861d0b3f4/F2.jpg", "media/caeba79eb00861d0b3f4/F3.jpg", "media/caeba79eb00861d0b3f4/F4.jpg", "media/caeba79eb00861d0b3f4/F5.jpg", "media/caeba79eb00861d0b3f4/F6.jpg", "media/caeba79eb00861d0b3f4/F7.jpg", "media/caeba79eb00861d0b3f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "09e3d253a59a43584aecc69c7b93312d438f21342e3718f4c0a0630a22270eca"}
{"qa_id": "MOSE__072e7b3f__object_world_motion_direction__01", "video_id": "MOSE__072e7b3f", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (cyclist in yellow shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/8cdba56bec5593ef35d9/F0.jpg", "media/8cdba56bec5593ef35d9/F1.jpg", "media/8cdba56bec5593ef35d9/F2.jpg", "media/8cdba56bec5593ef35d9/F3.jpg", "media/8cdba56bec5593ef35d9/F4.jpg", "media/8cdba56bec5593ef35d9/F5.jpg", "media/8cdba56bec5593ef35d9/F6.jpg", "media/8cdba56bec5593ef35d9/F7.jpg", "media/8cdba56bec5593ef35d9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (cyclist in yellow shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "375de3f7f8274c0f5076ab24466452281d1cf84c5dd3109cee10e19baae81611"}
{"qa_id": "MOSE__072e7b3f__camera_object_distance_change__02", "video_id": "MOSE__072e7b3f", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, did object A (cyclist in yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8cdba56bec5593ef35d9/F0.jpg", "media/8cdba56bec5593ef35d9/F1.jpg", "media/8cdba56bec5593ef35d9/F2.jpg", "media/8cdba56bec5593ef35d9/F3.jpg", "media/8cdba56bec5593ef35d9/F4.jpg", "media/8cdba56bec5593ef35d9/F5.jpg", "media/8cdba56bec5593ef35d9/F6.jpg", "media/8cdba56bec5593ef35d9/F7.jpg", "media/8cdba56bec5593ef35d9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (cyclist in yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36dc4d0724b3a30f085fdbc5cfd0ae013bdf660558aaf8160abddccbc3c68b4f"}
{"qa_id": "MOSE__0770ad03__object_world_motion_direction__00", "video_id": "MOSE__0770ad03", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object A (orange and yellow box truck) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "right", "backward"], "frame_paths": ["media/07594366e535f2742f7c/F0.jpg", "media/07594366e535f2742f7c/F1.jpg", "media/07594366e535f2742f7c/F2.jpg", "media/07594366e535f2742f7c/F3.jpg", "media/07594366e535f2742f7c/F4.jpg", "media/07594366e535f2742f7c/F5.jpg", "media/07594366e535f2742f7c/F6.jpg", "media/07594366e535f2742f7c/F7.jpg", "media/07594366e535f2742f7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object A (orange and yellow box truck) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6dc3991c290427448bede1c40a03cd2e92facd9396a728202d7ddc00e08244c6"}
{"qa_id": "MOSE__0770ad03__camera_object_distance_change__01", "video_id": "MOSE__0770ad03", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object B (dark grey suv) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d3108c14adb4e41193d3/F0.jpg", "media/d3108c14adb4e41193d3/F1.jpg", "media/d3108c14adb4e41193d3/F2.jpg", "media/d3108c14adb4e41193d3/F3.jpg", "media/d3108c14adb4e41193d3/F4.jpg", "media/d3108c14adb4e41193d3/F5.jpg", "media/d3108c14adb4e41193d3/F6.jpg", "media/d3108c14adb4e41193d3/F7.jpg", "media/d3108c14adb4e41193d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object B (dark grey suv) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "89c0eee54b4e287016f9411f3d3b4a1a8bdf7ecba9119b47b4d4a57dd9a8c57e"}
{"qa_id": "MOSE__0770ad03__object_pair_distance_change__02", "video_id": "MOSE__0770ad03", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did the 3D distance between object A (orange and yellow box truck) and object B (dark grey suv) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/cf1ec4c430d0e926cb22/F0.jpg", "media/cf1ec4c430d0e926cb22/F1.jpg", "media/cf1ec4c430d0e926cb22/F2.jpg", "media/cf1ec4c430d0e926cb22/F3.jpg", "media/cf1ec4c430d0e926cb22/F4.jpg", "media/cf1ec4c430d0e926cb22/F5.jpg", "media/cf1ec4c430d0e926cb22/F6.jpg", "media/cf1ec4c430d0e926cb22/F7.jpg", "media/cf1ec4c430d0e926cb22/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did the 3D distance between object A (orange and yellow box truck) and object B (dark grey suv) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca2466e9c652526baf7e914f7f8e024c172215407db9de4d0a02633ba2bdf49b"}
{"qa_id": "MOSE__0770ad03__object_travel_distance_comparison__03", "video_id": "MOSE__0770ad03", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange and yellow box truck) or B (dark grey suv)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/cf1ec4c430d0e926cb22/F0.jpg", "media/cf1ec4c430d0e926cb22/F1.jpg", "media/cf1ec4c430d0e926cb22/F2.jpg", "media/cf1ec4c430d0e926cb22/F3.jpg", "media/cf1ec4c430d0e926cb22/F4.jpg", "media/cf1ec4c430d0e926cb22/F5.jpg", "media/cf1ec4c430d0e926cb22/F6.jpg", "media/cf1ec4c430d0e926cb22/F7.jpg", "media/cf1ec4c430d0e926cb22/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange and yellow box truck) or B (dark grey suv)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e457905d905a97a4a8ec704ee084298f433dabb4afa9d8e3cf7e9376cd9a1c06"}
{"qa_id": "MOSE__07779415__camera_object_distance_change__00", "video_id": "MOSE__07779415", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/12afda81ae28df0c4908/F0.jpg", "media/12afda81ae28df0c4908/F1.jpg", "media/12afda81ae28df0c4908/F2.jpg", "media/12afda81ae28df0c4908/F3.jpg", "media/12afda81ae28df0c4908/F4.jpg", "media/12afda81ae28df0c4908/F5.jpg", "media/12afda81ae28df0c4908/F6.jpg", "media/12afda81ae28df0c4908/F7.jpg", "media/12afda81ae28df0c4908/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d14196ab348f738b1de8191864140678f608b39fd3a078303602bf48031a8072"}
{"qa_id": "MOSE__07b5d86c__object_world_motion_direction__01", "video_id": "MOSE__07b5d86c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object B (large elephant walking in the foreground) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/b50a62f1a171324cca10/F0.jpg", "media/b50a62f1a171324cca10/F1.jpg", "media/b50a62f1a171324cca10/F2.jpg", "media/b50a62f1a171324cca10/F3.jpg", "media/b50a62f1a171324cca10/F4.jpg", "media/b50a62f1a171324cca10/F5.jpg", "media/b50a62f1a171324cca10/F6.jpg", "media/b50a62f1a171324cca10/F7.jpg", "media/b50a62f1a171324cca10/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object B (large elephant walking in the foreground) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1d6891462359786ea1fb801dd131198408c2bce5445a0f4f701d8215a1375db"}
{"qa_id": "MOSE__07b5d86c__object_travel_distance_comparison__02", "video_id": "MOSE__07b5d86c", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant walking on the far right) or B (large elephant walking in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1339cba1ab2f129b69d2/F0.jpg", "media/1339cba1ab2f129b69d2/F1.jpg", "media/1339cba1ab2f129b69d2/F2.jpg", "media/1339cba1ab2f129b69d2/F3.jpg", "media/1339cba1ab2f129b69d2/F4.jpg", "media/1339cba1ab2f129b69d2/F5.jpg", "media/1339cba1ab2f129b69d2/F6.jpg", "media/1339cba1ab2f129b69d2/F7.jpg", "media/1339cba1ab2f129b69d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant walking on the far right) or B (large elephant walking in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0b264cdbc044839c5eda10bf915d53eea80a10f998533ca16e4bb3e26c4e8af"}
{"qa_id": "MOSE__08a5c662__object_world_motion_direction__00", "video_id": "MOSE__08a5c662", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (white cab concrete mixer truck with a large grey drum) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "up", "forward", "down"], "frame_paths": ["media/95d1ddf2a6fe8348a7ca/F0.jpg", "media/95d1ddf2a6fe8348a7ca/F1.jpg", "media/95d1ddf2a6fe8348a7ca/F2.jpg", "media/95d1ddf2a6fe8348a7ca/F3.jpg", "media/95d1ddf2a6fe8348a7ca/F4.jpg", "media/95d1ddf2a6fe8348a7ca/F5.jpg", "media/95d1ddf2a6fe8348a7ca/F6.jpg", "media/95d1ddf2a6fe8348a7ca/F7.jpg", "media/95d1ddf2a6fe8348a7ca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (white cab concrete mixer truck with a large grey drum) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "500636f7cf617a91ecb46f26e1cf7e68d12b480b3b268511605108dd10f6a2cd"}
{"qa_id": "MOSE__08a5c662__object_world_motion_direction__01", "video_id": "MOSE__08a5c662", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (grey sedan driving on the road) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/960998cadd7b4f02c9d0/F0.jpg", "media/960998cadd7b4f02c9d0/F1.jpg", "media/960998cadd7b4f02c9d0/F2.jpg", "media/960998cadd7b4f02c9d0/F3.jpg", "media/960998cadd7b4f02c9d0/F4.jpg", "media/960998cadd7b4f02c9d0/F5.jpg", "media/960998cadd7b4f02c9d0/F6.jpg", "media/960998cadd7b4f02c9d0/F7.jpg", "media/960998cadd7b4f02c9d0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (grey sedan driving on the road) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f35aa1054096e354a9b0ef3d65040223945da4f9a018d726e4fbbc21c5b2dd7"}
{"qa_id": "MOSE__08a5c662__object_pair_distance_change__02", "video_id": "MOSE__08a5c662", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did the 3D distance between object A (white cab concrete mixer truck with a large grey drum) and object B (grey sedan driving on the road) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/298c34760d89a35c0423/F0.jpg", "media/298c34760d89a35c0423/F1.jpg", "media/298c34760d89a35c0423/F2.jpg", "media/298c34760d89a35c0423/F3.jpg", "media/298c34760d89a35c0423/F4.jpg", "media/298c34760d89a35c0423/F5.jpg", "media/298c34760d89a35c0423/F6.jpg", "media/298c34760d89a35c0423/F7.jpg", "media/298c34760d89a35c0423/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (white cab concrete mixer truck with a large grey drum) and object B (grey sedan driving on the road) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a52edce3fa9dfcd723acde9289b5bf643ea3e84c8a60eea85d9221d137a14fb"}
{"qa_id": "MOSE__0a43a414__object_world_motion_direction__00", "video_id": "MOSE__0a43a414", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (dark brown cow with horns) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "forward", "right", "up"], "frame_paths": ["media/67778e01c1eb06d2b39a/F0.jpg", "media/67778e01c1eb06d2b39a/F1.jpg", "media/67778e01c1eb06d2b39a/F2.jpg", "media/67778e01c1eb06d2b39a/F3.jpg", "media/67778e01c1eb06d2b39a/F4.jpg", "media/67778e01c1eb06d2b39a/F5.jpg", "media/67778e01c1eb06d2b39a/F6.jpg", "media/67778e01c1eb06d2b39a/F7.jpg", "media/67778e01c1eb06d2b39a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (dark brown cow with horns) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. forward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ac928f69fb746a5a196b9dd26260682bec6977ae4e2457be613f5fe21ea9142d"}
{"qa_id": "MOSE__0af03282__object_world_motion_direction__00", "video_id": "MOSE__0af03282", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (motocross rider in black jersey with number 22 and orange helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "right", "backward"], "frame_paths": ["media/88b5dbea5091305618ef/F0.jpg", "media/88b5dbea5091305618ef/F1.jpg", "media/88b5dbea5091305618ef/F2.jpg", "media/88b5dbea5091305618ef/F3.jpg", "media/88b5dbea5091305618ef/F4.jpg", "media/88b5dbea5091305618ef/F5.jpg", "media/88b5dbea5091305618ef/F6.jpg", "media/88b5dbea5091305618ef/F7.jpg", "media/88b5dbea5091305618ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (motocross rider in black jersey with number 22 and orange helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f82931c93b906f1b9b5c5c013918ea5f721f1dc9005cb7587bec741d4d4d3457"}
{"qa_id": "MOSE__0af03282__object_pair_distance_change__02", "video_id": "MOSE__0af03282", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (motocross rider in black jersey with number 22 and orange helmet) and object B (motocross rider in black jersey with number 93 and white helmet) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/6732829432793d736f15/F0.jpg", "media/6732829432793d736f15/F1.jpg", "media/6732829432793d736f15/F2.jpg", "media/6732829432793d736f15/F3.jpg", "media/6732829432793d736f15/F4.jpg", "media/6732829432793d736f15/F5.jpg", "media/6732829432793d736f15/F6.jpg", "media/6732829432793d736f15/F7.jpg", "media/6732829432793d736f15/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (motocross rider in black jersey with number 22 and orange helmet) and object B (motocross rider in black jersey with number 93 and white helmet) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dba56c13cdf8a32196c989cdb8318f49051940dd15ab14579d4638cc26f78420"}
{"qa_id": "MOSE__0b5e5d29__object_pair_distance_change__00", "video_id": "MOSE__0b5e5d29", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (sheep with a black face and light woolly coat) and object B (sheep with a light face and woolly coat standing on the far right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/fbc3bbc9a3c9055a782e/F0.jpg", "media/fbc3bbc9a3c9055a782e/F1.jpg", "media/fbc3bbc9a3c9055a782e/F2.jpg", "media/fbc3bbc9a3c9055a782e/F3.jpg", "media/fbc3bbc9a3c9055a782e/F4.jpg", "media/fbc3bbc9a3c9055a782e/F5.jpg", "media/fbc3bbc9a3c9055a782e/F6.jpg", "media/fbc3bbc9a3c9055a782e/F7.jpg", "media/fbc3bbc9a3c9055a782e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (sheep with a black face and light woolly coat) and object B (sheep with a light face and woolly coat standing on the far right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2eb233baa99538831dec36151b95615bbff50c856524ece74d4a85f38880c90d"}
{"qa_id": "MOSE__0b5e5d29__object_travel_distance_comparison__01", "video_id": "MOSE__0b5e5d29", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with a black face and light woolly coat) or B (sheep with a light face and woolly coat standing on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/fbc3bbc9a3c9055a782e/F0.jpg", "media/fbc3bbc9a3c9055a782e/F1.jpg", "media/fbc3bbc9a3c9055a782e/F2.jpg", "media/fbc3bbc9a3c9055a782e/F3.jpg", "media/fbc3bbc9a3c9055a782e/F4.jpg", "media/fbc3bbc9a3c9055a782e/F5.jpg", "media/fbc3bbc9a3c9055a782e/F6.jpg", "media/fbc3bbc9a3c9055a782e/F7.jpg", "media/fbc3bbc9a3c9055a782e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with a black face and light woolly coat) or B (sheep with a light face and woolly coat standing on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e96726eae3a8f9982f8a98708ac0295db0f23f517e45442dcfc52306b40d78c0"}
{"qa_id": "MOSE__0ba9c5ad__camera_translation_direction__00", "video_id": "MOSE__0ba9c5ad", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "down", "right", "up", "left"], "frame_paths": ["media/485a79345a4549cbc9ac/F0.jpg", "media/485a79345a4549cbc9ac/F1.jpg", "media/485a79345a4549cbc9ac/F2.jpg", "media/485a79345a4549cbc9ac/F3.jpg", "media/485a79345a4549cbc9ac/F4.jpg", "media/485a79345a4549cbc9ac/F5.jpg", "media/485a79345a4549cbc9ac/F6.jpg", "media/485a79345a4549cbc9ac/F7.jpg", "media/485a79345a4549cbc9ac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. down\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e6adaef8a68500433160504593a3e673bae8944c907fca7ddb87605ee6f2088a"}
{"qa_id": "MOSE__0ba9c5ad__camera_object_distance_change__01", "video_id": "MOSE__0ba9c5ad", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (sailboat with a person wearing an orange life vest) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/76ccea2bbb64b18a12d3/F0.jpg", "media/76ccea2bbb64b18a12d3/F1.jpg", "media/76ccea2bbb64b18a12d3/F2.jpg", "media/76ccea2bbb64b18a12d3/F3.jpg", "media/76ccea2bbb64b18a12d3/F4.jpg", "media/76ccea2bbb64b18a12d3/F5.jpg", "media/76ccea2bbb64b18a12d3/F6.jpg", "media/76ccea2bbb64b18a12d3/F7.jpg", "media/76ccea2bbb64b18a12d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (sailboat with a person wearing an orange life vest) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1ff984c1af3a1c119b9de8140bb7b5400e53ed3cd8f74d8064d1c27f96d1f2fd"}
{"qa_id": "MOSE__0ba9c5ad__object_pair_distance_change__02", "video_id": "MOSE__0ba9c5ad", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (sailboat with a person wearing an orange life vest) and object B (sailboat with a person wearing a red life vest) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7e4850a09604f9f49aad/F0.jpg", "media/7e4850a09604f9f49aad/F1.jpg", "media/7e4850a09604f9f49aad/F2.jpg", "media/7e4850a09604f9f49aad/F3.jpg", "media/7e4850a09604f9f49aad/F4.jpg", "media/7e4850a09604f9f49aad/F5.jpg", "media/7e4850a09604f9f49aad/F6.jpg", "media/7e4850a09604f9f49aad/F7.jpg", "media/7e4850a09604f9f49aad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (sailboat with a person wearing an orange life vest) and object B (sailboat with a person wearing a red life vest) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6ccfc0f14ed911a56ddc6ad176afafb752f004fbe2026ae20eb18457602db7ca"}
{"qa_id": "MOSE__0d4feda2__camera_translation_direction__00", "video_id": "MOSE__0d4feda2", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "up", "left"], "frame_paths": ["media/1945010ceb7903546bcc/F0.jpg", "media/1945010ceb7903546bcc/F1.jpg", "media/1945010ceb7903546bcc/F2.jpg", "media/1945010ceb7903546bcc/F3.jpg", "media/1945010ceb7903546bcc/F4.jpg", "media/1945010ceb7903546bcc/F5.jpg", "media/1945010ceb7903546bcc/F6.jpg", "media/1945010ceb7903546bcc/F7.jpg", "media/1945010ceb7903546bcc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a13efe53ae5566ff95720d779cd0451135ecc8a4f0ae28da21ba0e4dac7c52c4"}
{"qa_id": "MOSE__0d4feda2__object_world_motion_direction__01", "video_id": "MOSE__0d4feda2", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (black rabbit partially obscured by green leaves) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "up", "left", "forward"], "frame_paths": ["media/a1914104faeab8345803/F0.jpg", "media/a1914104faeab8345803/F1.jpg", "media/a1914104faeab8345803/F2.jpg", "media/a1914104faeab8345803/F3.jpg", "media/a1914104faeab8345803/F4.jpg", "media/a1914104faeab8345803/F5.jpg", "media/a1914104faeab8345803/F6.jpg", "media/a1914104faeab8345803/F7.jpg", "media/a1914104faeab8345803/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (black rabbit partially obscured by green leaves) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66157c03bf67aa7d9cad2597db68b0d8e783869ef0ece904330aac86c16bdcf1"}
{"qa_id": "MOSE__0d4feda2__object_pair_distance_change__02", "video_id": "MOSE__0d4feda2", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (white rabbit with grey patches and brown ears) and object B (black rabbit partially obscured by green leaves) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/be73ac1e67aee869bfa3/F0.jpg", "media/be73ac1e67aee869bfa3/F1.jpg", "media/be73ac1e67aee869bfa3/F2.jpg", "media/be73ac1e67aee869bfa3/F3.jpg", "media/be73ac1e67aee869bfa3/F4.jpg", "media/be73ac1e67aee869bfa3/F5.jpg", "media/be73ac1e67aee869bfa3/F6.jpg", "media/be73ac1e67aee869bfa3/F7.jpg", "media/be73ac1e67aee869bfa3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (white rabbit with grey patches and brown ears) and object B (black rabbit partially obscured by green leaves) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ddca6c1a7331f2c2f110298d84c6b4bcf2da714c2de704d665657a59154946ca"}
{"qa_id": "MOSE__0d4feda2__object_travel_distance_comparison__03", "video_id": "MOSE__0d4feda2", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white rabbit with grey patches and brown ears) or B (black rabbit partially obscured by green leaves)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/be73ac1e67aee869bfa3/F0.jpg", "media/be73ac1e67aee869bfa3/F1.jpg", "media/be73ac1e67aee869bfa3/F2.jpg", "media/be73ac1e67aee869bfa3/F3.jpg", "media/be73ac1e67aee869bfa3/F4.jpg", "media/be73ac1e67aee869bfa3/F5.jpg", "media/be73ac1e67aee869bfa3/F6.jpg", "media/be73ac1e67aee869bfa3/F7.jpg", "media/be73ac1e67aee869bfa3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white rabbit with grey patches and brown ears) or B (black rabbit partially obscured by green leaves)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "627fe37f034aeaf2ca3efffd8da1a6c40fcef15b83da9ee05c3cf89432f09e9e"}
{"qa_id": "MOSE__0e05fabe__camera_object_distance_change__00", "video_id": "MOSE__0e05fabe", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (person in purple jacket and orange helmet riding a white scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/16ce4870375d61b4badf/F0.jpg", "media/16ce4870375d61b4badf/F1.jpg", "media/16ce4870375d61b4badf/F2.jpg", "media/16ce4870375d61b4badf/F3.jpg", "media/16ce4870375d61b4badf/F4.jpg", "media/16ce4870375d61b4badf/F5.jpg", "media/16ce4870375d61b4badf/F6.jpg", "media/16ce4870375d61b4badf/F7.jpg", "media/16ce4870375d61b4badf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (person in purple jacket and orange helmet riding a white scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "67ef297cae11030561f857fb4ad831bcf49995aaeea271f64b6695d8bb38ff00"}
{"qa_id": "MOSE__0e05fabe__object_pair_distance_change__01", "video_id": "MOSE__0e05fabe", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (person in purple jacket and orange helmet riding a white scooter) and object B (person in black jacket standing on the right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/ce16a6d9409116945ff9/F0.jpg", "media/ce16a6d9409116945ff9/F1.jpg", "media/ce16a6d9409116945ff9/F2.jpg", "media/ce16a6d9409116945ff9/F3.jpg", "media/ce16a6d9409116945ff9/F4.jpg", "media/ce16a6d9409116945ff9/F5.jpg", "media/ce16a6d9409116945ff9/F6.jpg", "media/ce16a6d9409116945ff9/F7.jpg", "media/ce16a6d9409116945ff9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (person in purple jacket and orange helmet riding a white scooter) and object B (person in black jacket standing on the right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "090c0dd0c47154da0e0304df510e9a585e008fc9ce2cd1698ce3af9a1a199e7d"}
{"qa_id": "MOSE__0e05fabe__object_travel_distance_comparison__02", "video_id": "MOSE__0e05fabe", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in purple jacket and orange helmet riding a white scooter) or B (person in black jacket standing on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/ce16a6d9409116945ff9/F0.jpg", "media/ce16a6d9409116945ff9/F1.jpg", "media/ce16a6d9409116945ff9/F2.jpg", "media/ce16a6d9409116945ff9/F3.jpg", "media/ce16a6d9409116945ff9/F4.jpg", "media/ce16a6d9409116945ff9/F5.jpg", "media/ce16a6d9409116945ff9/F6.jpg", "media/ce16a6d9409116945ff9/F7.jpg", "media/ce16a6d9409116945ff9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in purple jacket and orange helmet riding a white scooter) or B (person in black jacket standing on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d956bb219878f5b7efc11bb952be86f450be6a71ad4aea8733a662f6acf9f70"}
{"qa_id": "MOSE__0e7001dd__camera_translation_direction__00", "video_id": "MOSE__0e7001dd", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "down", "up", "forward"], "frame_paths": ["media/42aa6d658fdf52fea6f7/F0.jpg", "media/42aa6d658fdf52fea6f7/F1.jpg", "media/42aa6d658fdf52fea6f7/F2.jpg", "media/42aa6d658fdf52fea6f7/F3.jpg", "media/42aa6d658fdf52fea6f7/F4.jpg", "media/42aa6d658fdf52fea6f7/F5.jpg", "media/42aa6d658fdf52fea6f7/F6.jpg", "media/42aa6d658fdf52fea6f7/F7.jpg", "media/42aa6d658fdf52fea6f7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f050886f36f9dcfb54b688d9df5e11434c7f083eb22b7eee564b7932d515fcde"}
{"qa_id": "MOSE__0f3b1fcb__object_pair_distance_change__01", "video_id": "MOSE__0f3b1fcb", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did the 3D distance between object A (pink fish with a rounded body and distinct dorsal fin) and object B (pink fish with a rounded body, positioned below the other pink fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/171d0b93dab112440196/F0.jpg", "media/171d0b93dab112440196/F1.jpg", "media/171d0b93dab112440196/F2.jpg", "media/171d0b93dab112440196/F3.jpg", "media/171d0b93dab112440196/F4.jpg", "media/171d0b93dab112440196/F5.jpg", "media/171d0b93dab112440196/F6.jpg", "media/171d0b93dab112440196/F7.jpg", "media/171d0b93dab112440196/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did the 3D distance between object A (pink fish with a rounded body and distinct dorsal fin) and object B (pink fish with a rounded body, positioned below the other pink fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d59291553fcaeb2c01a2dfccfb2023b29738853b07b072e1b2d7b9369379b2b"}
{"qa_id": "MOSE__0f96e2d7__camera_view_direction_change__00", "video_id": "MOSE__0f96e2d7", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/c606bfcc452b9d05a667/F0.jpg", "media/c606bfcc452b9d05a667/F1.jpg", "media/c606bfcc452b9d05a667/F2.jpg", "media/c606bfcc452b9d05a667/F3.jpg", "media/c606bfcc452b9d05a667/F4.jpg", "media/c606bfcc452b9d05a667/F5.jpg", "media/c606bfcc452b9d05a667/F6.jpg", "media/c606bfcc452b9d05a667/F7.jpg", "media/c606bfcc452b9d05a667/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dba841b202b51c2333b93baedea94fb5af6539e50296e8c4b44f0e1807749f62"}
{"qa_id": "MOSE__0f96e2d7__object_world_motion_direction__01", "video_id": "MOSE__0f96e2d7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (large adult elephant walking in the herd) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "backward", "left", "down"], "frame_paths": ["media/1852a0b83ef6ad9a738d/F0.jpg", "media/1852a0b83ef6ad9a738d/F1.jpg", "media/1852a0b83ef6ad9a738d/F2.jpg", "media/1852a0b83ef6ad9a738d/F3.jpg", "media/1852a0b83ef6ad9a738d/F4.jpg", "media/1852a0b83ef6ad9a738d/F5.jpg", "media/1852a0b83ef6ad9a738d/F6.jpg", "media/1852a0b83ef6ad9a738d/F7.jpg", "media/1852a0b83ef6ad9a738d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (large adult elephant walking in the herd) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e293358d12262e7051247b1841f4fef1897fc947c9f97c9590ceab159ef4f9b"}
{"qa_id": "MOSE__0f96e2d7__camera_object_distance_change__02", "video_id": "MOSE__0f96e2d7", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did object A (large adult elephant walking in the herd) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/1852a0b83ef6ad9a738d/F0.jpg", "media/1852a0b83ef6ad9a738d/F1.jpg", "media/1852a0b83ef6ad9a738d/F2.jpg", "media/1852a0b83ef6ad9a738d/F3.jpg", "media/1852a0b83ef6ad9a738d/F4.jpg", "media/1852a0b83ef6ad9a738d/F5.jpg", "media/1852a0b83ef6ad9a738d/F6.jpg", "media/1852a0b83ef6ad9a738d/F7.jpg", "media/1852a0b83ef6ad9a738d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (large adult elephant walking in the herd) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d6615d2e3ba0ccd3f3e1eabb8679edb14db9b53f69a8c535a024e295efaa0d4"}
{"qa_id": "MOSE__0f96e2d7__object_world_motion_direction__03", "video_id": "MOSE__0f96e2d7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object B (small juvenile elephant walking near the adults) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "down", "left", "forward"], "frame_paths": ["media/40c7b032a204fdec6397/F0.jpg", "media/40c7b032a204fdec6397/F1.jpg", "media/40c7b032a204fdec6397/F2.jpg", "media/40c7b032a204fdec6397/F3.jpg", "media/40c7b032a204fdec6397/F4.jpg", "media/40c7b032a204fdec6397/F5.jpg", "media/40c7b032a204fdec6397/F6.jpg", "media/40c7b032a204fdec6397/F7.jpg", "media/40c7b032a204fdec6397/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object B (small juvenile elephant walking near the adults) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9204e1ce6f2cc29de95f94b4abb0b0b43385277c5afab3d85ba1d33e676f49de"}
{"qa_id": "MOSE__0f96e2d7__object_pair_distance_change__04", "video_id": "MOSE__0f96e2d7", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did the 3D distance between object A (large adult elephant walking in the herd) and object B (small juvenile elephant walking near the adults) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/7440bc53b19df55aec6d/F0.jpg", "media/7440bc53b19df55aec6d/F1.jpg", "media/7440bc53b19df55aec6d/F2.jpg", "media/7440bc53b19df55aec6d/F3.jpg", "media/7440bc53b19df55aec6d/F4.jpg", "media/7440bc53b19df55aec6d/F5.jpg", "media/7440bc53b19df55aec6d/F6.jpg", "media/7440bc53b19df55aec6d/F7.jpg", "media/7440bc53b19df55aec6d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1", "g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (large adult elephant walking in the herd) and object B (small juvenile elephant walking near the adults) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce6cbfe50e3b8f26b4c88b8d716730786eb46758a4b1bb475ca535fc3a59ac09"}
{"qa_id": "MOSE__100f6ecf__object_pair_distance_change__00", "video_id": "MOSE__100f6ecf", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did the 3D distance between object A (white and ginger cat lying on its back) and object B (ginger cat standing over the other cat) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/974d42c5085c4a58fcd2/F0.jpg", "media/974d42c5085c4a58fcd2/F1.jpg", "media/974d42c5085c4a58fcd2/F2.jpg", "media/974d42c5085c4a58fcd2/F3.jpg", "media/974d42c5085c4a58fcd2/F4.jpg", "media/974d42c5085c4a58fcd2/F5.jpg", "media/974d42c5085c4a58fcd2/F6.jpg", "media/974d42c5085c4a58fcd2/F7.jpg", "media/974d42c5085c4a58fcd2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (white and ginger cat lying on its back) and object B (ginger cat standing over the other cat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f74b3d7ca13a900c179059e1d3126b8c77fd0ee51697da420f8e9f97b23c98b"}
{"qa_id": "MOSE__100f6ecf__object_travel_distance_comparison__01", "video_id": "MOSE__100f6ecf", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white and ginger cat lying on its back) or B (ginger cat standing over the other cat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/974d42c5085c4a58fcd2/F0.jpg", "media/974d42c5085c4a58fcd2/F1.jpg", "media/974d42c5085c4a58fcd2/F2.jpg", "media/974d42c5085c4a58fcd2/F3.jpg", "media/974d42c5085c4a58fcd2/F4.jpg", "media/974d42c5085c4a58fcd2/F5.jpg", "media/974d42c5085c4a58fcd2/F6.jpg", "media/974d42c5085c4a58fcd2/F7.jpg", "media/974d42c5085c4a58fcd2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white and ginger cat lying on its back) or B (ginger cat standing over the other cat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "890a6234aaf6ec2fc004bb22c930bc5e090cf3866b8441583f061d74a66cb7d8"}
{"qa_id": "MOSE__111e0a5c__camera_view_direction_change__00", "video_id": "MOSE__111e0a5c", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/68189a45456d60571668/F0.jpg", "media/68189a45456d60571668/F1.jpg", "media/68189a45456d60571668/F2.jpg", "media/68189a45456d60571668/F3.jpg", "media/68189a45456d60571668/F4.jpg", "media/68189a45456d60571668/F5.jpg", "media/68189a45456d60571668/F6.jpg", "media/68189a45456d60571668/F7.jpg", "media/68189a45456d60571668/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "MOSE__111e0a5c__object_world_motion_direction__01", "video_id": "MOSE__111e0a5c", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (zebra walking to the right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "left", "forward", "backward", "up", "right"], "frame_paths": ["media/2d134d8ea8db525707f4/F0.jpg", "media/2d134d8ea8db525707f4/F1.jpg", "media/2d134d8ea8db525707f4/F2.jpg", "media/2d134d8ea8db525707f4/F3.jpg", "media/2d134d8ea8db525707f4/F4.jpg", "media/2d134d8ea8db525707f4/F5.jpg", "media/2d134d8ea8db525707f4/F6.jpg", "media/2d134d8ea8db525707f4/F7.jpg", "media/2d134d8ea8db525707f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (zebra walking to the right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. forward\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f22ab7044073044dbf05478db3f55045ce7e583e81cfefd0a054780f1aaa6f61"}
{"qa_id": "MOSE__111e0a5c__object_world_motion_direction__02", "video_id": "MOSE__111e0a5c", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (zebra walking to the right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "down", "right", "up", "left"], "frame_paths": ["media/884a63e9324b884fe976/F0.jpg", "media/884a63e9324b884fe976/F1.jpg", "media/884a63e9324b884fe976/F2.jpg", "media/884a63e9324b884fe976/F3.jpg", "media/884a63e9324b884fe976/F4.jpg", "media/884a63e9324b884fe976/F5.jpg", "media/884a63e9324b884fe976/F6.jpg", "media/884a63e9324b884fe976/F7.jpg", "media/884a63e9324b884fe976/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (zebra walking to the right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. down\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "572b139cfd3356d7451fca2c5bfd81443a02be52ea02bb87f26aaf2470ebfe39"}
{"qa_id": "MOSE__12a42fd9__camera_translation_direction__00", "video_id": "MOSE__12a42fd9", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "down", "up", "backward"], "frame_paths": ["media/73703e24634632a1d0da/F0.jpg", "media/73703e24634632a1d0da/F1.jpg", "media/73703e24634632a1d0da/F2.jpg", "media/73703e24634632a1d0da/F3.jpg", "media/73703e24634632a1d0da/F4.jpg", "media/73703e24634632a1d0da/F5.jpg", "media/73703e24634632a1d0da/F6.jpg", "media/73703e24634632a1d0da/F7.jpg", "media/73703e24634632a1d0da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d6dc901861683859c6c015e19f062f9e0f21ffc6b2d609e5c700732fa6b2ee7"}
{"qa_id": "MOSE__12a42fd9__camera_object_distance_change__01", "video_id": "MOSE__12a42fd9", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object A (man in grey plaid suit) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ffecf203babaf1e7f59e/F0.jpg", "media/ffecf203babaf1e7f59e/F1.jpg", "media/ffecf203babaf1e7f59e/F2.jpg", "media/ffecf203babaf1e7f59e/F3.jpg", "media/ffecf203babaf1e7f59e/F4.jpg", "media/ffecf203babaf1e7f59e/F5.jpg", "media/ffecf203babaf1e7f59e/F6.jpg", "media/ffecf203babaf1e7f59e/F7.jpg", "media/ffecf203babaf1e7f59e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (man in grey plaid suit) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "500066b5ecf45308014a142fee8dd9c815d366f753d904d973d65b733d4ce9d4"}
{"qa_id": "MOSE__12a42fd9__camera_object_distance_change__02", "video_id": "MOSE__12a42fd9", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object B (person in pink shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d71b29c93c441fdfe99d/F0.jpg", "media/d71b29c93c441fdfe99d/F1.jpg", "media/d71b29c93c441fdfe99d/F2.jpg", "media/d71b29c93c441fdfe99d/F3.jpg", "media/d71b29c93c441fdfe99d/F4.jpg", "media/d71b29c93c441fdfe99d/F5.jpg", "media/d71b29c93c441fdfe99d/F6.jpg", "media/d71b29c93c441fdfe99d/F7.jpg", "media/d71b29c93c441fdfe99d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (person in pink shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "56aa4662246e1a6052930f43788f939c464c283f691fc1da1af84eee59de8993"}
{"qa_id": "MOSE__12a5d0d7__camera_translation_direction__00", "video_id": "MOSE__12a5d0d7", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "left", "down", "forward"], "frame_paths": ["media/81a33787d56ace624642/F0.jpg", "media/81a33787d56ace624642/F1.jpg", "media/81a33787d56ace624642/F2.jpg", "media/81a33787d56ace624642/F3.jpg", "media/81a33787d56ace624642/F4.jpg", "media/81a33787d56ace624642/F5.jpg", "media/81a33787d56ace624642/F6.jpg", "media/81a33787d56ace624642/F7.jpg", "media/81a33787d56ace624642/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3945b25bd32eaa0dd704df1d2fdb5c1d9f134f8175dd7b9925435437b5bd196"}
{"qa_id": "MOSE__12e509c2__camera_object_distance_change__00", "video_id": "MOSE__12e509c2", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, did object A (white and tan dog with a fluffy tail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ca5c5e95f7d740c56ca6/F0.jpg", "media/ca5c5e95f7d740c56ca6/F1.jpg", "media/ca5c5e95f7d740c56ca6/F2.jpg", "media/ca5c5e95f7d740c56ca6/F3.jpg", "media/ca5c5e95f7d740c56ca6/F4.jpg", "media/ca5c5e95f7d740c56ca6/F5.jpg", "media/ca5c5e95f7d740c56ca6/F6.jpg", "media/ca5c5e95f7d740c56ca6/F7.jpg", "media/ca5c5e95f7d740c56ca6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (white and tan dog with a fluffy tail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "10b1400c1a4fb8bda092874008aa30f2db0a25361a421269b68c830a61b64845"}
{"qa_id": "MOSE__12fdc864__object_world_motion_direction__00", "video_id": "MOSE__12fdc864", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (zebra standing on the right side of the waterhole) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "down", "up", "forward"], "frame_paths": ["media/4d75cf47f2e94e8846e6/F0.jpg", "media/4d75cf47f2e94e8846e6/F1.jpg", "media/4d75cf47f2e94e8846e6/F2.jpg", "media/4d75cf47f2e94e8846e6/F3.jpg", "media/4d75cf47f2e94e8846e6/F4.jpg", "media/4d75cf47f2e94e8846e6/F5.jpg", "media/4d75cf47f2e94e8846e6/F6.jpg", "media/4d75cf47f2e94e8846e6/F7.jpg", "media/4d75cf47f2e94e8846e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (zebra standing on the right side of the waterhole) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a8cb456be4cdfb79c233f90504d4376fa187cf386c485647874e94f366f9977"}
{"qa_id": "MOSE__12fdc864__object_pair_distance_change__01", "video_id": "MOSE__12fdc864", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did the 3D distance between object A (zebra standing on the right side of the waterhole) and object B (zebra standing in the water on the left side) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/96064f79dfbbbaffbf1f/F0.jpg", "media/96064f79dfbbbaffbf1f/F1.jpg", "media/96064f79dfbbbaffbf1f/F2.jpg", "media/96064f79dfbbbaffbf1f/F3.jpg", "media/96064f79dfbbbaffbf1f/F4.jpg", "media/96064f79dfbbbaffbf1f/F5.jpg", "media/96064f79dfbbbaffbf1f/F6.jpg", "media/96064f79dfbbbaffbf1f/F7.jpg", "media/96064f79dfbbbaffbf1f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (zebra standing on the right side of the waterhole) and object B (zebra standing in the water on the left side) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "521b52e45953553db17cc5ea98dd66b3cd73ed4104ae57a508e21601cea433f0"}
{"qa_id": "MOSE__138c81fe__camera_view_direction_change__00", "video_id": "MOSE__138c81fe", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/6353a528a917c9ba277f/F0.jpg", "media/6353a528a917c9ba277f/F1.jpg", "media/6353a528a917c9ba277f/F2.jpg", "media/6353a528a917c9ba277f/F3.jpg", "media/6353a528a917c9ba277f/F4.jpg", "media/6353a528a917c9ba277f/F5.jpg", "media/6353a528a917c9ba277f/F6.jpg", "media/6353a528a917c9ba277f/F7.jpg", "media/6353a528a917c9ba277f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1912eec683eea1dfcf313a947a080c7cba700c52619bb2eddcbd50df4efaec8a"}
{"qa_id": "MOSE__13e6df75__camera_object_distance_change__00", "video_id": "MOSE__13e6df75", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b6182e9c8a2476ee7b1c/F0.jpg", "media/b6182e9c8a2476ee7b1c/F1.jpg", "media/b6182e9c8a2476ee7b1c/F2.jpg", "media/b6182e9c8a2476ee7b1c/F3.jpg", "media/b6182e9c8a2476ee7b1c/F4.jpg", "media/b6182e9c8a2476ee7b1c/F5.jpg", "media/b6182e9c8a2476ee7b1c/F6.jpg", "media/b6182e9c8a2476ee7b1c/F7.jpg", "media/b6182e9c8a2476ee7b1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a98a71dfdfe7e3125cab409c6e7893d1c9737d53cf68edaf9ce07f564472f906"}
{"qa_id": "MOSE__13f0606c__camera_view_direction_change__00", "video_id": "MOSE__13f0606c", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/9829845cf137087d390f/F0.jpg", "media/9829845cf137087d390f/F1.jpg", "media/9829845cf137087d390f/F2.jpg", "media/9829845cf137087d390f/F3.jpg", "media/9829845cf137087d390f/F4.jpg", "media/9829845cf137087d390f/F5.jpg", "media/9829845cf137087d390f/F6.jpg", "media/9829845cf137087d390f/F7.jpg", "media/9829845cf137087d390f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c70fea1841e5615b035b56b3517fe7bc5407eb2e748861736d937a6ed13d5c1"}
{"qa_id": "MOSE__13f0606c__object_world_motion_direction__01", "video_id": "MOSE__13f0606c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "up", "left", "backward"], "frame_paths": ["media/49f09b000678681b537f/F0.jpg", "media/49f09b000678681b537f/F1.jpg", "media/49f09b000678681b537f/F2.jpg", "media/49f09b000678681b537f/F3.jpg", "media/49f09b000678681b537f/F4.jpg", "media/49f09b000678681b537f/F5.jpg", "media/49f09b000678681b537f/F6.jpg", "media/49f09b000678681b537f/F7.jpg", "media/49f09b000678681b537f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. up\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a47244117b055449f4bd0fac54ce23c33caef30190f0bb52854d2e4164e9939a"}
{"qa_id": "MOSE__13f0606c__object_pair_distance_change__02", "video_id": "MOSE__13f0606c", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (black sedan) and object B (green shared bicycle) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/9dddc1f4fce3e0dafa36/F0.jpg", "media/9dddc1f4fce3e0dafa36/F1.jpg", "media/9dddc1f4fce3e0dafa36/F2.jpg", "media/9dddc1f4fce3e0dafa36/F3.jpg", "media/9dddc1f4fce3e0dafa36/F4.jpg", "media/9dddc1f4fce3e0dafa36/F5.jpg", "media/9dddc1f4fce3e0dafa36/F6.jpg", "media/9dddc1f4fce3e0dafa36/F7.jpg", "media/9dddc1f4fce3e0dafa36/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (black sedan) and object B (green shared bicycle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "155dc7cc9efa2812f56acd2fae3cd7d6161c70a67e631f515bb41bd3f0876eee"}
{"qa_id": "MOSE__149dbae2__camera_translation_direction__00", "video_id": "MOSE__149dbae2", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "forward", "backward"], "frame_paths": ["media/fe450b95657e14589611/F0.jpg", "media/fe450b95657e14589611/F1.jpg", "media/fe450b95657e14589611/F2.jpg", "media/fe450b95657e14589611/F3.jpg", "media/fe450b95657e14589611/F4.jpg", "media/fe450b95657e14589611/F5.jpg", "media/fe450b95657e14589611/F6.jpg", "media/fe450b95657e14589611/F7.jpg", "media/fe450b95657e14589611/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "47025472cf5999351f8fde7b7ebf1ad3e7a7112dec631d4734e9ce22431ffe69"}
{"qa_id": "MOSE__149dbae2__camera_view_direction_change__01", "video_id": "MOSE__149dbae2", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/fe450b95657e14589611/F0.jpg", "media/fe450b95657e14589611/F1.jpg", "media/fe450b95657e14589611/F2.jpg", "media/fe450b95657e14589611/F3.jpg", "media/fe450b95657e14589611/F4.jpg", "media/fe450b95657e14589611/F5.jpg", "media/fe450b95657e14589611/F6.jpg", "media/fe450b95657e14589611/F7.jpg", "media/fe450b95657e14589611/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c14dcd5f24642225f3bd8464554491ae5cd58ca8706517ba6816f0d0ae533c8"}
{"qa_id": "MOSE__149dbae2__camera_object_distance_change__02", "video_id": "MOSE__149dbae2", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object B (black horse standing on the left) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c092a3f7277a41db7ac8/F0.jpg", "media/c092a3f7277a41db7ac8/F1.jpg", "media/c092a3f7277a41db7ac8/F2.jpg", "media/c092a3f7277a41db7ac8/F3.jpg", "media/c092a3f7277a41db7ac8/F4.jpg", "media/c092a3f7277a41db7ac8/F5.jpg", "media/c092a3f7277a41db7ac8/F6.jpg", "media/c092a3f7277a41db7ac8/F7.jpg", "media/c092a3f7277a41db7ac8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object B (black horse standing on the left) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f61aa00940558dd55e047f27b32b24353e135da67d0c3ba37104c78bc3fd088c"}
{"qa_id": "MOSE__149dbae2__object_travel_distance_comparison__03", "video_id": "MOSE__149dbae2", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark brown horse standing on the right) or B (black horse standing on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/8b9233e20accbb7e452a/F0.jpg", "media/8b9233e20accbb7e452a/F1.jpg", "media/8b9233e20accbb7e452a/F2.jpg", "media/8b9233e20accbb7e452a/F3.jpg", "media/8b9233e20accbb7e452a/F4.jpg", "media/8b9233e20accbb7e452a/F5.jpg", "media/8b9233e20accbb7e452a/F6.jpg", "media/8b9233e20accbb7e452a/F7.jpg", "media/8b9233e20accbb7e452a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark brown horse standing on the right) or B (black horse standing on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f60ef038624cab0aa9ee364a26b0dc23fd5559ea9677d9376a642a635694039a"}
{"qa_id": "MOSE__158062aa__object_pair_distance_change__00", "video_id": "MOSE__158062aa", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (small red fish on the left) and object B (large red fish on the bottom right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4fab373d1388258a6b7b/F0.jpg", "media/4fab373d1388258a6b7b/F1.jpg", "media/4fab373d1388258a6b7b/F2.jpg", "media/4fab373d1388258a6b7b/F3.jpg", "media/4fab373d1388258a6b7b/F4.jpg", "media/4fab373d1388258a6b7b/F5.jpg", "media/4fab373d1388258a6b7b/F6.jpg", "media/4fab373d1388258a6b7b/F7.jpg", "media/4fab373d1388258a6b7b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (small red fish on the left) and object B (large red fish on the bottom right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d284050cd609ea4ef06c88305ae6b47adc2d199781904ccd23c1f44c4fb93843"}
{"qa_id": "MOSE__158062aa__object_travel_distance_comparison__01", "video_id": "MOSE__158062aa", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small red fish on the left) or B (large red fish on the bottom right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/4fab373d1388258a6b7b/F0.jpg", "media/4fab373d1388258a6b7b/F1.jpg", "media/4fab373d1388258a6b7b/F2.jpg", "media/4fab373d1388258a6b7b/F3.jpg", "media/4fab373d1388258a6b7b/F4.jpg", "media/4fab373d1388258a6b7b/F5.jpg", "media/4fab373d1388258a6b7b/F6.jpg", "media/4fab373d1388258a6b7b/F7.jpg", "media/4fab373d1388258a6b7b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small red fish on the left) or B (large red fish on the bottom right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16a229c6e8ab6c82975dde7220f98d2cdf73b9144707a9428b2ce0f22f888ab6"}
{"qa_id": "MOSE__15e9161e__object_world_motion_direction__00", "video_id": "MOSE__15e9161e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in orange shirt and white hat walking dogs) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/e010e25f5eb6f81823f2/F0.jpg", "media/e010e25f5eb6f81823f2/F1.jpg", "media/e010e25f5eb6f81823f2/F2.jpg", "media/e010e25f5eb6f81823f2/F3.jpg", "media/e010e25f5eb6f81823f2/F4.jpg", "media/e010e25f5eb6f81823f2/F5.jpg", "media/e010e25f5eb6f81823f2/F6.jpg", "media/e010e25f5eb6f81823f2/F7.jpg", "media/e010e25f5eb6f81823f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in orange shirt and white hat walking dogs) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c032e39e6c3d51122c8cac2063fd497d9b990ef09dc385b546e7a78a8ed3660"}
{"qa_id": "MOSE__17f66311__object_world_motion_direction__00", "video_id": "MOSE__17f66311", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a light-colored shirt riding a motorcycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "up", "forward", "left"], "frame_paths": ["media/891c302fe9d4fbe53228/F0.jpg", "media/891c302fe9d4fbe53228/F1.jpg", "media/891c302fe9d4fbe53228/F2.jpg", "media/891c302fe9d4fbe53228/F3.jpg", "media/891c302fe9d4fbe53228/F4.jpg", "media/891c302fe9d4fbe53228/F5.jpg", "media/891c302fe9d4fbe53228/F6.jpg", "media/891c302fe9d4fbe53228/F7.jpg", "media/891c302fe9d4fbe53228/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a light-colored shirt riding a motorcycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "172affcf8e857d1a1f7928ecd6741a4c5c0da61ba81944be915b3e37e82e56a4"}
{"qa_id": "MOSE__17f66311__camera_object_distance_change__01", "video_id": "MOSE__17f66311", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (motorcycle with a red taillight and visible front wheel) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/3f1cc94ea427c0a06a30/F0.jpg", "media/3f1cc94ea427c0a06a30/F1.jpg", "media/3f1cc94ea427c0a06a30/F2.jpg", "media/3f1cc94ea427c0a06a30/F3.jpg", "media/3f1cc94ea427c0a06a30/F4.jpg", "media/3f1cc94ea427c0a06a30/F5.jpg", "media/3f1cc94ea427c0a06a30/F6.jpg", "media/3f1cc94ea427c0a06a30/F7.jpg", "media/3f1cc94ea427c0a06a30/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (motorcycle with a red taillight and visible front wheel) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9749a8f10f840b5c524e1b05fdc590500967a88f02e21022b97b1b6910dd8793"}
{"qa_id": "MOSE__18840617__camera_object_distance_change__00", "video_id": "MOSE__18840617", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object A (person in light shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/060355124699488c2fc5/F0.jpg", "media/060355124699488c2fc5/F1.jpg", "media/060355124699488c2fc5/F2.jpg", "media/060355124699488c2fc5/F3.jpg", "media/060355124699488c2fc5/F4.jpg", "media/060355124699488c2fc5/F5.jpg", "media/060355124699488c2fc5/F6.jpg", "media/060355124699488c2fc5/F7.jpg", "media/060355124699488c2fc5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person in light shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "663a6d401ad3ade32dcbb3f289662717cbc329a1b3924515d000e911559f3267"}
{"qa_id": "MOSE__199995b5__camera_view_direction_change__00", "video_id": "MOSE__199995b5", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/fe7c14a208f8d4432bae/F0.jpg", "media/fe7c14a208f8d4432bae/F1.jpg", "media/fe7c14a208f8d4432bae/F2.jpg", "media/fe7c14a208f8d4432bae/F3.jpg", "media/fe7c14a208f8d4432bae/F4.jpg", "media/fe7c14a208f8d4432bae/F5.jpg", "media/fe7c14a208f8d4432bae/F6.jpg", "media/fe7c14a208f8d4432bae/F7.jpg", "media/fe7c14a208f8d4432bae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "37e670c6cbe146b110034b978d60f4e932b21c96881fab92eb5e568c9a6249e2"}
{"qa_id": "MOSE__199995b5__object_world_motion_direction__01", "video_id": "MOSE__199995b5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (giraffe standing on the right side of the group) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "down", "right"], "frame_paths": ["media/02030d23f3d54e7cdb4b/F0.jpg", "media/02030d23f3d54e7cdb4b/F1.jpg", "media/02030d23f3d54e7cdb4b/F2.jpg", "media/02030d23f3d54e7cdb4b/F3.jpg", "media/02030d23f3d54e7cdb4b/F4.jpg", "media/02030d23f3d54e7cdb4b/F5.jpg", "media/02030d23f3d54e7cdb4b/F6.jpg", "media/02030d23f3d54e7cdb4b/F7.jpg", "media/02030d23f3d54e7cdb4b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (giraffe standing on the right side of the group) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "39869f852c3c698b75d237f7ed35651e54f4b393008a76b91cc1eb3beaa3af03"}
{"qa_id": "MOSE__1a6a731f__camera_translation_direction__00", "video_id": "MOSE__1a6a731f", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "right", "up", "backward"], "frame_paths": ["media/e9637480e6bec04f5c80/F0.jpg", "media/e9637480e6bec04f5c80/F1.jpg", "media/e9637480e6bec04f5c80/F2.jpg", "media/e9637480e6bec04f5c80/F3.jpg", "media/e9637480e6bec04f5c80/F4.jpg", "media/e9637480e6bec04f5c80/F5.jpg", "media/e9637480e6bec04f5c80/F6.jpg", "media/e9637480e6bec04f5c80/F7.jpg", "media/e9637480e6bec04f5c80/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cabddf462310cfa568d4af5e3cb6acd1718d8ad8439f77141ae15bcb2ed160f7"}
{"qa_id": "MOSE__1b7eada8__camera_view_direction_change__00", "video_id": "MOSE__1b7eada8", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/f8b75317193f8038e59e/F0.jpg", "media/f8b75317193f8038e59e/F1.jpg", "media/f8b75317193f8038e59e/F2.jpg", "media/f8b75317193f8038e59e/F3.jpg", "media/f8b75317193f8038e59e/F4.jpg", "media/f8b75317193f8038e59e/F5.jpg", "media/f8b75317193f8038e59e/F6.jpg", "media/f8b75317193f8038e59e/F7.jpg", "media/f8b75317193f8038e59e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9ba2cc51b29405424f7d48f9d823ee5cd59f07f17efd65cd4610ac1ec100a34d"}
{"qa_id": "MOSE__1c28e081__object_world_motion_direction__00", "video_id": "MOSE__1c28e081", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (black compact suv with white decals on the side) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/f005dfd831d19bb14146/F0.jpg", "media/f005dfd831d19bb14146/F1.jpg", "media/f005dfd831d19bb14146/F2.jpg", "media/f005dfd831d19bb14146/F3.jpg", "media/f005dfd831d19bb14146/F4.jpg", "media/f005dfd831d19bb14146/F5.jpg", "media/f005dfd831d19bb14146/F6.jpg", "media/f005dfd831d19bb14146/F7.jpg", "media/f005dfd831d19bb14146/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (black compact suv with white decals on the side) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0303a106d80233396ef309805c8cc84ed1c4bbee051430d8cb210455eb7374d9"}
{"qa_id": "MOSE__1c28e081__camera_object_distance_change__01", "video_id": "MOSE__1c28e081", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (black compact suv with white decals on the side) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f005dfd831d19bb14146/F0.jpg", "media/f005dfd831d19bb14146/F1.jpg", "media/f005dfd831d19bb14146/F2.jpg", "media/f005dfd831d19bb14146/F3.jpg", "media/f005dfd831d19bb14146/F4.jpg", "media/f005dfd831d19bb14146/F5.jpg", "media/f005dfd831d19bb14146/F6.jpg", "media/f005dfd831d19bb14146/F7.jpg", "media/f005dfd831d19bb14146/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (black compact suv with white decals on the side) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bbe194b648dc8af02f3193274b0986ec17b541539a3d5773da5b3eb71d6e1031"}
{"qa_id": "MOSE__1ca2531a__camera_object_distance_change__00", "video_id": "MOSE__1ca2531a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F6, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/15fe46ff13a3b5007b02/F0.jpg", "media/15fe46ff13a3b5007b02/F1.jpg", "media/15fe46ff13a3b5007b02/F2.jpg", "media/15fe46ff13a3b5007b02/F3.jpg", "media/15fe46ff13a3b5007b02/F4.jpg", "media/15fe46ff13a3b5007b02/F5.jpg", "media/15fe46ff13a3b5007b02/F6.jpg", "media/15fe46ff13a3b5007b02/F7.jpg", "media/15fe46ff13a3b5007b02/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "58f67fd150e4a820d3bf4f47dbded0ef635b267e2f4fd96963af55725c6503e8"}
{"qa_id": "MOSE__1d29944c__camera_view_direction_change__00", "video_id": "MOSE__1d29944c", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/5e45fb3004e600812e59/F0.jpg", "media/5e45fb3004e600812e59/F1.jpg", "media/5e45fb3004e600812e59/F2.jpg", "media/5e45fb3004e600812e59/F3.jpg", "media/5e45fb3004e600812e59/F4.jpg", "media/5e45fb3004e600812e59/F5.jpg", "media/5e45fb3004e600812e59/F6.jpg", "media/5e45fb3004e600812e59/F7.jpg", "media/5e45fb3004e600812e59/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bf827dc04ca477cc7349bcde6eae98c84940bf0f8cd3c76ee27c4cb2c52194e9"}
{"qa_id": "MOSE__1d29944c__object_pair_distance_change__01", "video_id": "MOSE__1d29944c", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did the 3D distance between object A (person in black jacket riding a scooter) and object B (person in black jacket and dark pants walking) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/eeb45196b25543547be9/F0.jpg", "media/eeb45196b25543547be9/F1.jpg", "media/eeb45196b25543547be9/F2.jpg", "media/eeb45196b25543547be9/F3.jpg", "media/eeb45196b25543547be9/F4.jpg", "media/eeb45196b25543547be9/F5.jpg", "media/eeb45196b25543547be9/F6.jpg", "media/eeb45196b25543547be9/F7.jpg", "media/eeb45196b25543547be9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (person in black jacket riding a scooter) and object B (person in black jacket and dark pants walking) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "157df5ee1c4ca5c6370597cfed04d78f4275852a973f84c34e04a50e96bc3509"}
{"qa_id": "MOSE__1dba51af__object_pair_distance_change__00", "video_id": "MOSE__1dba51af", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (dark brown duck swimming in the water) and object B (dark brown duck swimming in the water) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/ef7c9d0beb39343aaf78/F0.jpg", "media/ef7c9d0beb39343aaf78/F1.jpg", "media/ef7c9d0beb39343aaf78/F2.jpg", "media/ef7c9d0beb39343aaf78/F3.jpg", "media/ef7c9d0beb39343aaf78/F4.jpg", "media/ef7c9d0beb39343aaf78/F5.jpg", "media/ef7c9d0beb39343aaf78/F6.jpg", "media/ef7c9d0beb39343aaf78/F7.jpg", "media/ef7c9d0beb39343aaf78/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (dark brown duck swimming in the water) and object B (dark brown duck swimming in the water) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5ebfe936621b6f866b42460aed45ece285c2aaac49f5a06caaf53c74c0f69eb9"}
{"qa_id": "MOSE__1dbca5cc__object_pair_distance_change__00", "video_id": "MOSE__1dbca5cc", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (grey kitten standing in the center) and object B (grey kitten sitting on the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b02647beca8a98d3abe4/F0.jpg", "media/b02647beca8a98d3abe4/F1.jpg", "media/b02647beca8a98d3abe4/F2.jpg", "media/b02647beca8a98d3abe4/F3.jpg", "media/b02647beca8a98d3abe4/F4.jpg", "media/b02647beca8a98d3abe4/F5.jpg", "media/b02647beca8a98d3abe4/F6.jpg", "media/b02647beca8a98d3abe4/F7.jpg", "media/b02647beca8a98d3abe4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (grey kitten standing in the center) and object B (grey kitten sitting on the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c27f0efabb8dbcc2cbba356e7ffd763c474031b137de424e92f64f72baf7cb0"}
{"qa_id": "MOSE__1e356390__camera_object_distance_change__00", "video_id": "MOSE__1e356390", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F6 to F8, did object A (person in orange high-visibility workwear) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b2164971c77f7555ff07/F0.jpg", "media/b2164971c77f7555ff07/F1.jpg", "media/b2164971c77f7555ff07/F2.jpg", "media/b2164971c77f7555ff07/F3.jpg", "media/b2164971c77f7555ff07/F4.jpg", "media/b2164971c77f7555ff07/F5.jpg", "media/b2164971c77f7555ff07/F6.jpg", "media/b2164971c77f7555ff07/F7.jpg", "media/b2164971c77f7555ff07/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (person in orange high-visibility workwear) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8cd4007671f5c1a0f61c27c33a16b43cc11ad425baba8c1dfde76be1d1983aea"}
{"qa_id": "MOSE__1e95ef5e__camera_translation_direction__00", "video_id": "MOSE__1e95ef5e", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "down", "right"], "frame_paths": ["media/f8c5e89416170d5d943b/F0.jpg", "media/f8c5e89416170d5d943b/F1.jpg", "media/f8c5e89416170d5d943b/F2.jpg", "media/f8c5e89416170d5d943b/F3.jpg", "media/f8c5e89416170d5d943b/F4.jpg", "media/f8c5e89416170d5d943b/F5.jpg", "media/f8c5e89416170d5d943b/F6.jpg", "media/f8c5e89416170d5d943b/F7.jpg", "media/f8c5e89416170d5d943b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f18cbbc3d84b0d6a9ebff27accd8818a91b91189ab9194ab0aae4d4ea0ffb34"}
{"qa_id": "MOSE__1e95ef5e__object_pair_distance_change__03", "video_id": "MOSE__1e95ef5e", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did the 3D distance between object A (brown duck on the far right) and object B (brown duck on the far left) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7525b3e0aae06889490a/F0.jpg", "media/7525b3e0aae06889490a/F1.jpg", "media/7525b3e0aae06889490a/F2.jpg", "media/7525b3e0aae06889490a/F3.jpg", "media/7525b3e0aae06889490a/F4.jpg", "media/7525b3e0aae06889490a/F5.jpg", "media/7525b3e0aae06889490a/F6.jpg", "media/7525b3e0aae06889490a/F7.jpg", "media/7525b3e0aae06889490a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (brown duck on the far right) and object B (brown duck on the far left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91323f0190e8f61d382affead6d17148d9ddd8b281ce22215770d04e26623603"}
{"qa_id": "MOSE__1f220f98__object_world_motion_direction__00", "video_id": "MOSE__1f220f98", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (black cat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/5840e31d91566a1b2945/F0.jpg", "media/5840e31d91566a1b2945/F1.jpg", "media/5840e31d91566a1b2945/F2.jpg", "media/5840e31d91566a1b2945/F3.jpg", "media/5840e31d91566a1b2945/F4.jpg", "media/5840e31d91566a1b2945/F5.jpg", "media/5840e31d91566a1b2945/F6.jpg", "media/5840e31d91566a1b2945/F7.jpg", "media/5840e31d91566a1b2945/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (black cat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "adf2559babda9a812c24a9bf8e6a9dd1fc3dea499ac10c751441dbc017fde2ae"}
{"qa_id": "MOSE__1fab1d1c__object_world_motion_direction__00", "video_id": "MOSE__1fab1d1c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (large grey elephant walking away from the camera) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "left", "forward", "down"], "frame_paths": ["media/0b24785ff890d1819a2e/F0.jpg", "media/0b24785ff890d1819a2e/F1.jpg", "media/0b24785ff890d1819a2e/F2.jpg", "media/0b24785ff890d1819a2e/F3.jpg", "media/0b24785ff890d1819a2e/F4.jpg", "media/0b24785ff890d1819a2e/F5.jpg", "media/0b24785ff890d1819a2e/F6.jpg", "media/0b24785ff890d1819a2e/F7.jpg", "media/0b24785ff890d1819a2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (large grey elephant walking away from the camera) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0be6eb0cd393f5e5f3ac0b831f2f3b32edd6a19a0862b63e89b7586940d2c9c"}
{"qa_id": "MOSE__1fab1d1c__camera_object_distance_change__01", "video_id": "MOSE__1fab1d1c", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (large grey elephant walking away from the camera) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/0b24785ff890d1819a2e/F0.jpg", "media/0b24785ff890d1819a2e/F1.jpg", "media/0b24785ff890d1819a2e/F2.jpg", "media/0b24785ff890d1819a2e/F3.jpg", "media/0b24785ff890d1819a2e/F4.jpg", "media/0b24785ff890d1819a2e/F5.jpg", "media/0b24785ff890d1819a2e/F6.jpg", "media/0b24785ff890d1819a2e/F7.jpg", "media/0b24785ff890d1819a2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (large grey elephant walking away from the camera) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c12aacff3a41dd5ae98bc341955fb9a171d3f7ad875aaf7c88d087786aa34bbb"}
{"qa_id": "MOSE__1fc0e6a8__object_world_motion_direction__00", "video_id": "MOSE__1fc0e6a8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (zebra grazing in the center of the group) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "left", "up"], "frame_paths": ["media/3fec66e60347b0f7751a/F0.jpg", "media/3fec66e60347b0f7751a/F1.jpg", "media/3fec66e60347b0f7751a/F2.jpg", "media/3fec66e60347b0f7751a/F3.jpg", "media/3fec66e60347b0f7751a/F4.jpg", "media/3fec66e60347b0f7751a/F5.jpg", "media/3fec66e60347b0f7751a/F6.jpg", "media/3fec66e60347b0f7751a/F7.jpg", "media/3fec66e60347b0f7751a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (zebra grazing in the center of the group) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "226e332c208abaa7c7c05034c14bb612e1c0717177fc1f9c7064792ba6516ba2"}
{"qa_id": "MOSE__1fc2314b__camera_object_distance_change__00", "video_id": "MOSE__1fc2314b", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object B (sheep with light cream-colored wool) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b468276cfc4ab2ef203f/F0.jpg", "media/b468276cfc4ab2ef203f/F1.jpg", "media/b468276cfc4ab2ef203f/F2.jpg", "media/b468276cfc4ab2ef203f/F3.jpg", "media/b468276cfc4ab2ef203f/F4.jpg", "media/b468276cfc4ab2ef203f/F5.jpg", "media/b468276cfc4ab2ef203f/F6.jpg", "media/b468276cfc4ab2ef203f/F7.jpg", "media/b468276cfc4ab2ef203f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object B (sheep with light cream-colored wool) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4a428db37449ca46a002037de8a5b4c5532feefbd9a205cc31ec76d68aea16df"}
{"qa_id": "MOSE__1fc2314b__object_pair_distance_change__01", "video_id": "MOSE__1fc2314b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did the 3D distance between object A (sheep with brownish wool and a white face) and object B (sheep with light cream-colored wool) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/b7cd0635018e91af0464/F0.jpg", "media/b7cd0635018e91af0464/F1.jpg", "media/b7cd0635018e91af0464/F2.jpg", "media/b7cd0635018e91af0464/F3.jpg", "media/b7cd0635018e91af0464/F4.jpg", "media/b7cd0635018e91af0464/F5.jpg", "media/b7cd0635018e91af0464/F6.jpg", "media/b7cd0635018e91af0464/F7.jpg", "media/b7cd0635018e91af0464/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (sheep with brownish wool and a white face) and object B (sheep with light cream-colored wool) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7007d055273c49ac322ffebb80417a4b408bb95ed017b005ebdec3db173315fc"}
{"qa_id": "MOSE__1fc2314b__object_travel_distance_comparison__02", "video_id": "MOSE__1fc2314b", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with brownish wool and a white face) or B (sheep with light cream-colored wool)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/b7cd0635018e91af0464/F0.jpg", "media/b7cd0635018e91af0464/F1.jpg", "media/b7cd0635018e91af0464/F2.jpg", "media/b7cd0635018e91af0464/F3.jpg", "media/b7cd0635018e91af0464/F4.jpg", "media/b7cd0635018e91af0464/F5.jpg", "media/b7cd0635018e91af0464/F6.jpg", "media/b7cd0635018e91af0464/F7.jpg", "media/b7cd0635018e91af0464/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with brownish wool and a white face) or B (sheep with light cream-colored wool)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ada3de0933840925b40bb242a38bd078e715113c17d192086227e56425b97592"}
{"qa_id": "MOSE__1feae28e__object_world_motion_direction__00", "video_id": "MOSE__1feae28e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (large elephant standing on the right side of the enclosure) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "left", "up", "backward", "right", "forward"], "frame_paths": ["media/628df8067a5a20522b45/F0.jpg", "media/628df8067a5a20522b45/F1.jpg", "media/628df8067a5a20522b45/F2.jpg", "media/628df8067a5a20522b45/F3.jpg", "media/628df8067a5a20522b45/F4.jpg", "media/628df8067a5a20522b45/F5.jpg", "media/628df8067a5a20522b45/F6.jpg", "media/628df8067a5a20522b45/F7.jpg", "media/628df8067a5a20522b45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (large elephant standing on the right side of the enclosure) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. up\nD. backward\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3ea0e65a5b29b15d4648b9716b2397a8032d25f2a17daee7d2cdae00317ccaf2"}
{"qa_id": "MOSE__1feae28e__object_world_motion_direction__01", "video_id": "MOSE__1feae28e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (young elephant near the water's edge) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "backward", "forward"], "frame_paths": ["media/c29d3d0865ab833b8577/F0.jpg", "media/c29d3d0865ab833b8577/F1.jpg", "media/c29d3d0865ab833b8577/F2.jpg", "media/c29d3d0865ab833b8577/F3.jpg", "media/c29d3d0865ab833b8577/F4.jpg", "media/c29d3d0865ab833b8577/F5.jpg", "media/c29d3d0865ab833b8577/F6.jpg", "media/c29d3d0865ab833b8577/F7.jpg", "media/c29d3d0865ab833b8577/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (young elephant near the water's edge) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "baa4a975842c154aed2bb72bfb4322708c30129323a18d2dd628165c96ab14c8"}
{"qa_id": "MOSE__1feae28e__object_travel_distance_comparison__02", "video_id": "MOSE__1feae28e", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant standing on the right side of the enclosure) or B (young elephant near the water's edge)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/7e39abf69959b1d65e4a/F0.jpg", "media/7e39abf69959b1d65e4a/F1.jpg", "media/7e39abf69959b1d65e4a/F2.jpg", "media/7e39abf69959b1d65e4a/F3.jpg", "media/7e39abf69959b1d65e4a/F4.jpg", "media/7e39abf69959b1d65e4a/F5.jpg", "media/7e39abf69959b1d65e4a/F6.jpg", "media/7e39abf69959b1d65e4a/F7.jpg", "media/7e39abf69959b1d65e4a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large elephant standing on the right side of the enclosure) or B (young elephant near the water's edge)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "98911b0142073f33933b1aa448f9cdfd1334a1d96d97f0a977fec686a1bc66e5"}
{"qa_id": "MOSE__2080824a__object_world_motion_direction__00", "video_id": "MOSE__2080824a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "right", "left", "down"], "frame_paths": ["media/6d674692bd146c490f51/F0.jpg", "media/6d674692bd146c490f51/F1.jpg", "media/6d674692bd146c490f51/F2.jpg", "media/6d674692bd146c490f51/F3.jpg", "media/6d674692bd146c490f51/F4.jpg", "media/6d674692bd146c490f51/F5.jpg", "media/6d674692bd146c490f51/F6.jpg", "media/6d674692bd146c490f51/F7.jpg", "media/6d674692bd146c490f51/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d35da114d83357b1e603b9751e29509964a53db14c7fe51f2b4ae7f6d2c247dd"}
{"qa_id": "MOSE__2080824a__camera_object_distance_change__01", "video_id": "MOSE__2080824a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/6d674692bd146c490f51/F0.jpg", "media/6d674692bd146c490f51/F1.jpg", "media/6d674692bd146c490f51/F2.jpg", "media/6d674692bd146c490f51/F3.jpg", "media/6d674692bd146c490f51/F4.jpg", "media/6d674692bd146c490f51/F5.jpg", "media/6d674692bd146c490f51/F6.jpg", "media/6d674692bd146c490f51/F7.jpg", "media/6d674692bd146c490f51/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "239394c25871ab58a9e3153ace43484f26f30c7e0b0804baef3fa7000d4a3c08"}
{"qa_id": "MOSE__21f363b3__object_world_motion_direction__00", "video_id": "MOSE__21f363b3", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "up", "left", "backward"], "frame_paths": ["media/1bc6b315cc16bf266da7/F0.jpg", "media/1bc6b315cc16bf266da7/F1.jpg", "media/1bc6b315cc16bf266da7/F2.jpg", "media/1bc6b315cc16bf266da7/F3.jpg", "media/1bc6b315cc16bf266da7/F4.jpg", "media/1bc6b315cc16bf266da7/F5.jpg", "media/1bc6b315cc16bf266da7/F6.jpg", "media/1bc6b315cc16bf266da7/F7.jpg", "media/1bc6b315cc16bf266da7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. up\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6ef650db6b5f24f631702e2cc6ccaf1a27ed71767d42563086662abbfb976ca1"}
{"qa_id": "MOSE__22167a50__camera_view_direction_change__00", "video_id": "MOSE__22167a50", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/9d0d7c700703d3b8671a/F0.jpg", "media/9d0d7c700703d3b8671a/F1.jpg", "media/9d0d7c700703d3b8671a/F2.jpg", "media/9d0d7c700703d3b8671a/F3.jpg", "media/9d0d7c700703d3b8671a/F4.jpg", "media/9d0d7c700703d3b8671a/F5.jpg", "media/9d0d7c700703d3b8671a/F6.jpg", "media/9d0d7c700703d3b8671a/F7.jpg", "media/9d0d7c700703d3b8671a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ffaa2ca65e20ef4a1352c570aa9ab3a1b6d2ba632cff886aa32a0d994008c4a8"}
{"qa_id": "MOSE__228a0234__camera_object_distance_change__00", "video_id": "MOSE__228a0234", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object A (dark horse with a rider in a sandy arena) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/da5985476bf6a7c4e738/F0.jpg", "media/da5985476bf6a7c4e738/F1.jpg", "media/da5985476bf6a7c4e738/F2.jpg", "media/da5985476bf6a7c4e738/F3.jpg", "media/da5985476bf6a7c4e738/F4.jpg", "media/da5985476bf6a7c4e738/F5.jpg", "media/da5985476bf6a7c4e738/F6.jpg", "media/da5985476bf6a7c4e738/F7.jpg", "media/da5985476bf6a7c4e738/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (dark horse with a rider in a sandy arena) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c8f33fade092b9f843ffd0df7d2a642c1170c5b5a04f7b116ea241b36f6d1014"}
{"qa_id": "MOSE__230ddb91__object_world_motion_direction__00", "video_id": "MOSE__230ddb91", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (zebra in the foreground with distinct black and white stripes) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "right", "left", "down"], "frame_paths": ["media/37c279bd2621ec03068e/F0.jpg", "media/37c279bd2621ec03068e/F1.jpg", "media/37c279bd2621ec03068e/F2.jpg", "media/37c279bd2621ec03068e/F3.jpg", "media/37c279bd2621ec03068e/F4.jpg", "media/37c279bd2621ec03068e/F5.jpg", "media/37c279bd2621ec03068e/F6.jpg", "media/37c279bd2621ec03068e/F7.jpg", "media/37c279bd2621ec03068e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (zebra in the foreground with distinct black and white stripes) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dca0d9e922e1f2a75bd4ddde9dda442e411dcde240a9b02be788899ebaa98244"}
{"qa_id": "MOSE__230ddb91__object_pair_distance_change__01", "video_id": "MOSE__230ddb91", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (zebra in the foreground with distinct black and white stripes) and object B (zebra standing to the right of the foreground zebra) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b3c0a544b72b45125b5b/F0.jpg", "media/b3c0a544b72b45125b5b/F1.jpg", "media/b3c0a544b72b45125b5b/F2.jpg", "media/b3c0a544b72b45125b5b/F3.jpg", "media/b3c0a544b72b45125b5b/F4.jpg", "media/b3c0a544b72b45125b5b/F5.jpg", "media/b3c0a544b72b45125b5b/F6.jpg", "media/b3c0a544b72b45125b5b/F7.jpg", "media/b3c0a544b72b45125b5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (zebra in the foreground with distinct black and white stripes) and object B (zebra standing to the right of the foreground zebra) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a6d73f5ea2491975216400545b8a5073ba00f47663630e13143f6b2b126562f"}
{"qa_id": "MOSE__23b9a3ea__camera_object_distance_change__00", "video_id": "MOSE__23b9a3ea", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object A (striped fish with vertical black and white bands) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/e6f568fb5503877a297f/F0.jpg", "media/e6f568fb5503877a297f/F1.jpg", "media/e6f568fb5503877a297f/F2.jpg", "media/e6f568fb5503877a297f/F3.jpg", "media/e6f568fb5503877a297f/F4.jpg", "media/e6f568fb5503877a297f/F5.jpg", "media/e6f568fb5503877a297f/F6.jpg", "media/e6f568fb5503877a297f/F7.jpg", "media/e6f568fb5503877a297f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (striped fish with vertical black and white bands) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4c2288562b721e80ff4ad29928e34de0d03c85154fc59454394b5c571b2da7d9"}
{"qa_id": "MOSE__23b9a3ea__object_pair_distance_change__03", "video_id": "MOSE__23b9a3ea", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (striped fish with vertical black and white bands) and object B (striped fish with vertical black and white bands) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/33e5d89039918e220113/F0.jpg", "media/33e5d89039918e220113/F1.jpg", "media/33e5d89039918e220113/F2.jpg", "media/33e5d89039918e220113/F3.jpg", "media/33e5d89039918e220113/F4.jpg", "media/33e5d89039918e220113/F5.jpg", "media/33e5d89039918e220113/F6.jpg", "media/33e5d89039918e220113/F7.jpg", "media/33e5d89039918e220113/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (striped fish with vertical black and white bands) and object B (striped fish with vertical black and white bands) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "92afa35bf398459d86b9a7a4a4445ecf9ca3b4666c6fcb327820ddf5c9ddd7a3"}
{"qa_id": "MOSE__23e046fb__object_world_motion_direction__00", "video_id": "MOSE__23e046fb", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (zebra with distinct black and white stripes, positioned near a giraffe) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "left", "up"], "frame_paths": ["media/344bbddda8f1b1d337ed/F0.jpg", "media/344bbddda8f1b1d337ed/F1.jpg", "media/344bbddda8f1b1d337ed/F2.jpg", "media/344bbddda8f1b1d337ed/F3.jpg", "media/344bbddda8f1b1d337ed/F4.jpg", "media/344bbddda8f1b1d337ed/F5.jpg", "media/344bbddda8f1b1d337ed/F6.jpg", "media/344bbddda8f1b1d337ed/F7.jpg", "media/344bbddda8f1b1d337ed/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (zebra with distinct black and white stripes, positioned near a giraffe) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f247f1ce5158d75effcf0d76384053f57a6dbb514c3780767d59776a8f8fb1cb"}
{"qa_id": "MOSE__23e046fb__object_world_motion_direction__01", "video_id": "MOSE__23e046fb", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (zebra with distinct black and white stripes, positioned to the right of the other zebra) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "down", "up", "backward"], "frame_paths": ["media/4e58fdb25228d7a8b5cb/F0.jpg", "media/4e58fdb25228d7a8b5cb/F1.jpg", "media/4e58fdb25228d7a8b5cb/F2.jpg", "media/4e58fdb25228d7a8b5cb/F3.jpg", "media/4e58fdb25228d7a8b5cb/F4.jpg", "media/4e58fdb25228d7a8b5cb/F5.jpg", "media/4e58fdb25228d7a8b5cb/F6.jpg", "media/4e58fdb25228d7a8b5cb/F7.jpg", "media/4e58fdb25228d7a8b5cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (zebra with distinct black and white stripes, positioned to the right of the other zebra) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7d05cd934cd1ac8c179b2dc2a1a428290e4f242192837b6f4299cffea0e3355"}
{"qa_id": "MOSE__2463609b__object_pair_distance_change__00", "video_id": "MOSE__2463609b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (tiger cub with orange and black stripes) and object B (tiger cub with orange and black stripes) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7f63a86b6578c66459ad/F0.jpg", "media/7f63a86b6578c66459ad/F1.jpg", "media/7f63a86b6578c66459ad/F2.jpg", "media/7f63a86b6578c66459ad/F3.jpg", "media/7f63a86b6578c66459ad/F4.jpg", "media/7f63a86b6578c66459ad/F5.jpg", "media/7f63a86b6578c66459ad/F6.jpg", "media/7f63a86b6578c66459ad/F7.jpg", "media/7f63a86b6578c66459ad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (tiger cub with orange and black stripes) and object B (tiger cub with orange and black stripes) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2aa5b63d9b61f98209f053d751f659f04845484faa09c61ac3749e7564d0d3b"}
{"qa_id": "MOSE__25153e58__camera_view_direction_change__00", "video_id": "MOSE__25153e58", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/347bc8aada091b217de1/F0.jpg", "media/347bc8aada091b217de1/F1.jpg", "media/347bc8aada091b217de1/F2.jpg", "media/347bc8aada091b217de1/F3.jpg", "media/347bc8aada091b217de1/F4.jpg", "media/347bc8aada091b217de1/F5.jpg", "media/347bc8aada091b217de1/F6.jpg", "media/347bc8aada091b217de1/F7.jpg", "media/347bc8aada091b217de1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6236fb847960198c195f1d2dadf7c8193f905bb374b3dd0e53dd6f4b8f2f18f4"}
{"qa_id": "MOSE__265a69f4__camera_translation_direction__00", "video_id": "MOSE__265a69f4", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "left", "up"], "frame_paths": ["media/49612509f0b7faed7cc1/F0.jpg", "media/49612509f0b7faed7cc1/F1.jpg", "media/49612509f0b7faed7cc1/F2.jpg", "media/49612509f0b7faed7cc1/F3.jpg", "media/49612509f0b7faed7cc1/F4.jpg", "media/49612509f0b7faed7cc1/F5.jpg", "media/49612509f0b7faed7cc1/F6.jpg", "media/49612509f0b7faed7cc1/F7.jpg", "media/49612509f0b7faed7cc1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0aac5b72915c2863ea59676e43883c8387c37d9286c6dc0485188426a4c3971"}
{"qa_id": "MOSE__265a69f4__camera_view_direction_change__01", "video_id": "MOSE__265a69f4", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/49612509f0b7faed7cc1/F0.jpg", "media/49612509f0b7faed7cc1/F1.jpg", "media/49612509f0b7faed7cc1/F2.jpg", "media/49612509f0b7faed7cc1/F3.jpg", "media/49612509f0b7faed7cc1/F4.jpg", "media/49612509f0b7faed7cc1/F5.jpg", "media/49612509f0b7faed7cc1/F6.jpg", "media/49612509f0b7faed7cc1/F7.jpg", "media/49612509f0b7faed7cc1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48b0dbe5136add4561fdd3af171b3ed657216b06d587528f3932aa08151751c1"}
{"qa_id": "MOSE__2697f864__object_world_motion_direction__00", "video_id": "MOSE__2697f864", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person wearing a long orange coat and black pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "left", "forward", "right"], "frame_paths": ["media/042ece0df451d890369c/F0.jpg", "media/042ece0df451d890369c/F1.jpg", "media/042ece0df451d890369c/F2.jpg", "media/042ece0df451d890369c/F3.jpg", "media/042ece0df451d890369c/F4.jpg", "media/042ece0df451d890369c/F5.jpg", "media/042ece0df451d890369c/F6.jpg", "media/042ece0df451d890369c/F7.jpg", "media/042ece0df451d890369c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person wearing a long orange coat and black pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. left\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb415e66fd46f1e93a3c2819216c55da85126af570532839158663ea8a632bd7"}
{"qa_id": "MOSE__26ed56e6__object_world_motion_direction__00", "video_id": "MOSE__26ed56e6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (black car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/aa463a4612bb32082705/F0.jpg", "media/aa463a4612bb32082705/F1.jpg", "media/aa463a4612bb32082705/F2.jpg", "media/aa463a4612bb32082705/F3.jpg", "media/aa463a4612bb32082705/F4.jpg", "media/aa463a4612bb32082705/F5.jpg", "media/aa463a4612bb32082705/F6.jpg", "media/aa463a4612bb32082705/F7.jpg", "media/aa463a4612bb32082705/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (black car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb4ba5adc0c0280d64d58a5ed02201f4ca485e623e4d6769c93a9ea6dafeb937"}
{"qa_id": "MOSE__2744f35a__object_world_motion_direction__00", "video_id": "MOSE__2744f35a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (white cow) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/72bff4102e27bf2c0013/F0.jpg", "media/72bff4102e27bf2c0013/F1.jpg", "media/72bff4102e27bf2c0013/F2.jpg", "media/72bff4102e27bf2c0013/F3.jpg", "media/72bff4102e27bf2c0013/F4.jpg", "media/72bff4102e27bf2c0013/F5.jpg", "media/72bff4102e27bf2c0013/F6.jpg", "media/72bff4102e27bf2c0013/F7.jpg", "media/72bff4102e27bf2c0013/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (white cow) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2ae7e2732d1bdf883392150b25b847d75ae3e6bb48fc12ce61b299ddebb981a7"}
{"qa_id": "MOSE__277ad883__camera_view_direction_change__00", "video_id": "MOSE__277ad883", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/41de0c729ce857c138f3/F0.jpg", "media/41de0c729ce857c138f3/F1.jpg", "media/41de0c729ce857c138f3/F2.jpg", "media/41de0c729ce857c138f3/F3.jpg", "media/41de0c729ce857c138f3/F4.jpg", "media/41de0c729ce857c138f3/F5.jpg", "media/41de0c729ce857c138f3/F6.jpg", "media/41de0c729ce857c138f3/F7.jpg", "media/41de0c729ce857c138f3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5c8895c3358f98ed2d150b5b42d1d416d1985526e0470d0788e41e16d2795ef"}
{"qa_id": "MOSE__2883ae72__object_pair_distance_change__00", "video_id": "MOSE__2883ae72", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (person in blue uniform holding a panda) and object B (panda cub on the ground) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/04ba94ee4777b84e10fe/F0.jpg", "media/04ba94ee4777b84e10fe/F1.jpg", "media/04ba94ee4777b84e10fe/F2.jpg", "media/04ba94ee4777b84e10fe/F3.jpg", "media/04ba94ee4777b84e10fe/F4.jpg", "media/04ba94ee4777b84e10fe/F5.jpg", "media/04ba94ee4777b84e10fe/F6.jpg", "media/04ba94ee4777b84e10fe/F7.jpg", "media/04ba94ee4777b84e10fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in blue uniform holding a panda) and object B (panda cub on the ground) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "63f08c61f934b0dc6b0efd75998140cc38fd1090cabe177037f3319edf78f816"}
{"qa_id": "MOSE__2883ae72__object_travel_distance_comparison__01", "video_id": "MOSE__2883ae72", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue uniform holding a panda) or B (panda cub on the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/04ba94ee4777b84e10fe/F0.jpg", "media/04ba94ee4777b84e10fe/F1.jpg", "media/04ba94ee4777b84e10fe/F2.jpg", "media/04ba94ee4777b84e10fe/F3.jpg", "media/04ba94ee4777b84e10fe/F4.jpg", "media/04ba94ee4777b84e10fe/F5.jpg", "media/04ba94ee4777b84e10fe/F6.jpg", "media/04ba94ee4777b84e10fe/F7.jpg", "media/04ba94ee4777b84e10fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue uniform holding a panda) or B (panda cub on the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73443ee0a56310f0662c8d36cd869aa57fd8b28b80ceee516809c44a9f5b5a03"}
{"qa_id": "MOSE__28980657__object_world_motion_direction__00", "video_id": "MOSE__28980657", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange tabby cat with a green collar) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "up", "forward", "down"], "frame_paths": ["media/e196bcf94f94ba1a17ec/F0.jpg", "media/e196bcf94f94ba1a17ec/F1.jpg", "media/e196bcf94f94ba1a17ec/F2.jpg", "media/e196bcf94f94ba1a17ec/F3.jpg", "media/e196bcf94f94ba1a17ec/F4.jpg", "media/e196bcf94f94ba1a17ec/F5.jpg", "media/e196bcf94f94ba1a17ec/F6.jpg", "media/e196bcf94f94ba1a17ec/F7.jpg", "media/e196bcf94f94ba1a17ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange tabby cat with a green collar) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3549f270753959f70d9ac55ad6cc518fdc5656772eefbff6412a0e6430fd13ca"}
{"qa_id": "MOSE__291d4799__object_world_motion_direction__00", "video_id": "MOSE__291d4799", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object B (zebra running through the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "left", "down", "right"], "frame_paths": ["media/35e1ef8671881158ce0b/F0.jpg", "media/35e1ef8671881158ce0b/F1.jpg", "media/35e1ef8671881158ce0b/F2.jpg", "media/35e1ef8671881158ce0b/F3.jpg", "media/35e1ef8671881158ce0b/F4.jpg", "media/35e1ef8671881158ce0b/F5.jpg", "media/35e1ef8671881158ce0b/F6.jpg", "media/35e1ef8671881158ce0b/F7.jpg", "media/35e1ef8671881158ce0b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object B (zebra running through the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "304b1c8b106e2d3f26cd47cc6ead3a0f35f45d0c2820f1bff3db99a6e6953b21"}
{"qa_id": "MOSE__2ab9ba0f__object_world_motion_direction__00", "video_id": "MOSE__2ab9ba0f", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (grey remote-controlled fighter jet model with twin vertical tails and a star decal on the wing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "forward", "down", "right"], "frame_paths": ["media/f34efab48704d930d45a/F0.jpg", "media/f34efab48704d930d45a/F1.jpg", "media/f34efab48704d930d45a/F2.jpg", "media/f34efab48704d930d45a/F3.jpg", "media/f34efab48704d930d45a/F4.jpg", "media/f34efab48704d930d45a/F5.jpg", "media/f34efab48704d930d45a/F6.jpg", "media/f34efab48704d930d45a/F7.jpg", "media/f34efab48704d930d45a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (grey remote-controlled fighter jet model with twin vertical tails and a star decal on the wing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. forward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b3191b74f65b7c214f1588506de8263bfdc6877759d549873541d6d181db5988"}
{"qa_id": "MOSE__2bd4e845__object_pair_distance_change__00", "video_id": "MOSE__2bd4e845", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (pink fish with red speckles) and object B (dark reddish-brown fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/629c5a39cec1fbec8149/F0.jpg", "media/629c5a39cec1fbec8149/F1.jpg", "media/629c5a39cec1fbec8149/F2.jpg", "media/629c5a39cec1fbec8149/F3.jpg", "media/629c5a39cec1fbec8149/F4.jpg", "media/629c5a39cec1fbec8149/F5.jpg", "media/629c5a39cec1fbec8149/F6.jpg", "media/629c5a39cec1fbec8149/F7.jpg", "media/629c5a39cec1fbec8149/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (pink fish with red speckles) and object B (dark reddish-brown fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "833782d7b8a691a9225c339ab0246b7772d846b870933866834d133647cf6057"}
{"qa_id": "MOSE__2d3b0320__camera_translation_direction__00", "video_id": "MOSE__2d3b0320", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "right", "backward", "forward"], "frame_paths": ["media/a04240f4b035ff43865f/F0.jpg", "media/a04240f4b035ff43865f/F1.jpg", "media/a04240f4b035ff43865f/F2.jpg", "media/a04240f4b035ff43865f/F3.jpg", "media/a04240f4b035ff43865f/F4.jpg", "media/a04240f4b035ff43865f/F5.jpg", "media/a04240f4b035ff43865f/F6.jpg", "media/a04240f4b035ff43865f/F7.jpg", "media/a04240f4b035ff43865f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "71001478883fc8a20f62e5b5ab7477abe878f09b6255279e3b3d4f6df881a4a3"}
{"qa_id": "MOSE__2d403c59__object_world_motion_direction__00", "video_id": "MOSE__2d403c59", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (zebra standing near a giraffe) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "forward", "right", "left"], "frame_paths": ["media/a9de29ec6a7bb1a59c77/F0.jpg", "media/a9de29ec6a7bb1a59c77/F1.jpg", "media/a9de29ec6a7bb1a59c77/F2.jpg", "media/a9de29ec6a7bb1a59c77/F3.jpg", "media/a9de29ec6a7bb1a59c77/F4.jpg", "media/a9de29ec6a7bb1a59c77/F5.jpg", "media/a9de29ec6a7bb1a59c77/F6.jpg", "media/a9de29ec6a7bb1a59c77/F7.jpg", "media/a9de29ec6a7bb1a59c77/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (zebra standing near a giraffe) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b27b4efaf55fecd478d1adada0160b16f67417502c7ffeba8dc2e21590df9372"}
{"qa_id": "MOSE__2d8d7d35__camera_translation_direction__00", "video_id": "MOSE__2d8d7d35", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/c37a5c7bc947a8b9f4ca/F0.jpg", "media/c37a5c7bc947a8b9f4ca/F1.jpg", "media/c37a5c7bc947a8b9f4ca/F2.jpg", "media/c37a5c7bc947a8b9f4ca/F3.jpg", "media/c37a5c7bc947a8b9f4ca/F4.jpg", "media/c37a5c7bc947a8b9f4ca/F5.jpg", "media/c37a5c7bc947a8b9f4ca/F6.jpg", "media/c37a5c7bc947a8b9f4ca/F7.jpg", "media/c37a5c7bc947a8b9f4ca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "58ead3af71162682c16700df846e1634b0c87aa93b7423ce20eaae292910fa9a"}
{"qa_id": "MOSE__2d8d7d35__object_world_motion_direction__01", "video_id": "MOSE__2d8d7d35", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (dark bird with mottled brown and black feathers perched on a wooden dowel) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "left", "forward", "backward"], "frame_paths": ["media/8d1acacfb6d6fa41e09a/F0.jpg", "media/8d1acacfb6d6fa41e09a/F1.jpg", "media/8d1acacfb6d6fa41e09a/F2.jpg", "media/8d1acacfb6d6fa41e09a/F3.jpg", "media/8d1acacfb6d6fa41e09a/F4.jpg", "media/8d1acacfb6d6fa41e09a/F5.jpg", "media/8d1acacfb6d6fa41e09a/F6.jpg", "media/8d1acacfb6d6fa41e09a/F7.jpg", "media/8d1acacfb6d6fa41e09a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (dark bird with mottled brown and black feathers perched on a wooden dowel) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8db1000fbe38677dcd94a6cc2a71441bc32cef27cbbb0fdc070cba91e0cea7ae"}
{"qa_id": "MOSE__2d9101b8__camera_view_direction_change__00", "video_id": "MOSE__2d9101b8", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/0a77b0ca7eb6d6328d4b/F0.jpg", "media/0a77b0ca7eb6d6328d4b/F1.jpg", "media/0a77b0ca7eb6d6328d4b/F2.jpg", "media/0a77b0ca7eb6d6328d4b/F3.jpg", "media/0a77b0ca7eb6d6328d4b/F4.jpg", "media/0a77b0ca7eb6d6328d4b/F5.jpg", "media/0a77b0ca7eb6d6328d4b/F6.jpg", "media/0a77b0ca7eb6d6328d4b/F7.jpg", "media/0a77b0ca7eb6d6328d4b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "MOSE__2e85e790__camera_object_distance_change__00", "video_id": "MOSE__2e85e790", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object A (white sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9a598e6f86ffde2fa954/F0.jpg", "media/9a598e6f86ffde2fa954/F1.jpg", "media/9a598e6f86ffde2fa954/F2.jpg", "media/9a598e6f86ffde2fa954/F3.jpg", "media/9a598e6f86ffde2fa954/F4.jpg", "media/9a598e6f86ffde2fa954/F5.jpg", "media/9a598e6f86ffde2fa954/F6.jpg", "media/9a598e6f86ffde2fa954/F7.jpg", "media/9a598e6f86ffde2fa954/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (white sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a74b3e4253d8334cf1b430176e3eb3768415721c764c7b6769a6e5580d7415b"}
{"qa_id": "MOSE__2ea465de__camera_object_distance_change__00", "video_id": "MOSE__2ea465de", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c9c6edab109eea1843be/F0.jpg", "media/c9c6edab109eea1843be/F1.jpg", "media/c9c6edab109eea1843be/F2.jpg", "media/c9c6edab109eea1843be/F3.jpg", "media/c9c6edab109eea1843be/F4.jpg", "media/c9c6edab109eea1843be/F5.jpg", "media/c9c6edab109eea1843be/F6.jpg", "media/c9c6edab109eea1843be/F7.jpg", "media/c9c6edab109eea1843be/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8529dd51c4c8aa2f893ccfe556171fbe693d1f22acff7590d6d9baa65c4013d7"}
{"qa_id": "MOSE__2ea465de__object_pair_distance_change__01", "video_id": "MOSE__2ea465de", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (black sedan) and object B (man in black suit and white shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8435ee8552b4579f55f6/F0.jpg", "media/8435ee8552b4579f55f6/F1.jpg", "media/8435ee8552b4579f55f6/F2.jpg", "media/8435ee8552b4579f55f6/F3.jpg", "media/8435ee8552b4579f55f6/F4.jpg", "media/8435ee8552b4579f55f6/F5.jpg", "media/8435ee8552b4579f55f6/F6.jpg", "media/8435ee8552b4579f55f6/F7.jpg", "media/8435ee8552b4579f55f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (black sedan) and object B (man in black suit and white shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ffebea4381b202909759cf8fffddce4c36e9f018f34710a3df71ef6a2325fecf"}
{"qa_id": "MOSE__2ea465de__object_travel_distance_comparison__02", "video_id": "MOSE__2ea465de", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black sedan) or B (man in black suit and white shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/8435ee8552b4579f55f6/F0.jpg", "media/8435ee8552b4579f55f6/F1.jpg", "media/8435ee8552b4579f55f6/F2.jpg", "media/8435ee8552b4579f55f6/F3.jpg", "media/8435ee8552b4579f55f6/F4.jpg", "media/8435ee8552b4579f55f6/F5.jpg", "media/8435ee8552b4579f55f6/F6.jpg", "media/8435ee8552b4579f55f6/F7.jpg", "media/8435ee8552b4579f55f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black sedan) or B (man in black suit and white shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64b3ba79fca82a8232ab3ae1f08ab53096c2be0feae3be22f0ab14bda88b4be8"}
{"qa_id": "MOSE__2eb73c7c__camera_object_distance_change__00", "video_id": "MOSE__2eb73c7c", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object A (light brown chicken) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9a2b3068ef9a87549568/F0.jpg", "media/9a2b3068ef9a87549568/F1.jpg", "media/9a2b3068ef9a87549568/F2.jpg", "media/9a2b3068ef9a87549568/F3.jpg", "media/9a2b3068ef9a87549568/F4.jpg", "media/9a2b3068ef9a87549568/F5.jpg", "media/9a2b3068ef9a87549568/F6.jpg", "media/9a2b3068ef9a87549568/F7.jpg", "media/9a2b3068ef9a87549568/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (light brown chicken) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b8b45d57cd5219543b61dfefbd733513352c1d6a7b9ad3859edd392a66d48235"}
{"qa_id": "MOSE__2f477d05__camera_translation_direction__00", "video_id": "MOSE__2f477d05", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "backward", "up", "down"], "frame_paths": ["media/3d36dfbb83be13bfe9c4/F0.jpg", "media/3d36dfbb83be13bfe9c4/F1.jpg", "media/3d36dfbb83be13bfe9c4/F2.jpg", "media/3d36dfbb83be13bfe9c4/F3.jpg", "media/3d36dfbb83be13bfe9c4/F4.jpg", "media/3d36dfbb83be13bfe9c4/F5.jpg", "media/3d36dfbb83be13bfe9c4/F6.jpg", "media/3d36dfbb83be13bfe9c4/F7.jpg", "media/3d36dfbb83be13bfe9c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbf5731b86e22a420382e557c3dcd6dc9a21b9ca6bbcfd71abb650742a89374f"}
{"qa_id": "MOSE__2f477d05__camera_view_direction_change__01", "video_id": "MOSE__2f477d05", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/3d36dfbb83be13bfe9c4/F0.jpg", "media/3d36dfbb83be13bfe9c4/F1.jpg", "media/3d36dfbb83be13bfe9c4/F2.jpg", "media/3d36dfbb83be13bfe9c4/F3.jpg", "media/3d36dfbb83be13bfe9c4/F4.jpg", "media/3d36dfbb83be13bfe9c4/F5.jpg", "media/3d36dfbb83be13bfe9c4/F6.jpg", "media/3d36dfbb83be13bfe9c4/F7.jpg", "media/3d36dfbb83be13bfe9c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b307af0f2e559a886d19371bab5d46b899223979121bd89c34941ef2b1afc8d6"}
{"qa_id": "MOSE__30d1cb6b__camera_object_distance_change__00", "video_id": "MOSE__30d1cb6b", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (sheep with a black face and light wool) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/1eb1e96454a38ae36f52/F0.jpg", "media/1eb1e96454a38ae36f52/F1.jpg", "media/1eb1e96454a38ae36f52/F2.jpg", "media/1eb1e96454a38ae36f52/F3.jpg", "media/1eb1e96454a38ae36f52/F4.jpg", "media/1eb1e96454a38ae36f52/F5.jpg", "media/1eb1e96454a38ae36f52/F6.jpg", "media/1eb1e96454a38ae36f52/F7.jpg", "media/1eb1e96454a38ae36f52/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (sheep with a black face and light wool) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "690226b86d09eb06b363f68b7cd73af8ba85fcbfa31df3f7d543ff9cffb53b9c"}
{"qa_id": "MOSE__30d1cb6b__object_pair_distance_change__01", "video_id": "MOSE__30d1cb6b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did the 3D distance between object A (sheep with a black face and light wool) and object B (sheep with a black face and light wool) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/9e0d26287fb922b2bc3d/F0.jpg", "media/9e0d26287fb922b2bc3d/F1.jpg", "media/9e0d26287fb922b2bc3d/F2.jpg", "media/9e0d26287fb922b2bc3d/F3.jpg", "media/9e0d26287fb922b2bc3d/F4.jpg", "media/9e0d26287fb922b2bc3d/F5.jpg", "media/9e0d26287fb922b2bc3d/F6.jpg", "media/9e0d26287fb922b2bc3d/F7.jpg", "media/9e0d26287fb922b2bc3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (sheep with a black face and light wool) and object B (sheep with a black face and light wool) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b919f72f2be912fbf2ca1d26cc59ddfc7741e1058e1feaaf4c596617391c5e79"}
{"qa_id": "MOSE__3142014e__camera_object_distance_change__00", "video_id": "MOSE__3142014e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (white and purple SBS Transit double-decker bus) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ae798618d667a44e9695/F0.jpg", "media/ae798618d667a44e9695/F1.jpg", "media/ae798618d667a44e9695/F2.jpg", "media/ae798618d667a44e9695/F3.jpg", "media/ae798618d667a44e9695/F4.jpg", "media/ae798618d667a44e9695/F5.jpg", "media/ae798618d667a44e9695/F6.jpg", "media/ae798618d667a44e9695/F7.jpg", "media/ae798618d667a44e9695/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (white and purple SBS Transit double-decker bus) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "769181a14babebe60449152683f361cad3505799dbf7cb97c1de09f5566ae391"}
{"qa_id": "MOSE__3142014e__object_pair_distance_change__01", "video_id": "MOSE__3142014e", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (white and purple SBS Transit double-decker bus) and object B (silver hatchback car) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/a0893598cd596a148239/F0.jpg", "media/a0893598cd596a148239/F1.jpg", "media/a0893598cd596a148239/F2.jpg", "media/a0893598cd596a148239/F3.jpg", "media/a0893598cd596a148239/F4.jpg", "media/a0893598cd596a148239/F5.jpg", "media/a0893598cd596a148239/F6.jpg", "media/a0893598cd596a148239/F7.jpg", "media/a0893598cd596a148239/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (white and purple SBS Transit double-decker bus) and object B (silver hatchback car) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2e3e58e6c6645034440bab702ad00b584cfa6f2ebb045b545dafc3555131bfda"}
{"qa_id": "MOSE__3142014e__object_travel_distance_comparison__02", "video_id": "MOSE__3142014e", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white and purple SBS Transit double-decker bus) or B (silver hatchback car)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/a0893598cd596a148239/F0.jpg", "media/a0893598cd596a148239/F1.jpg", "media/a0893598cd596a148239/F2.jpg", "media/a0893598cd596a148239/F3.jpg", "media/a0893598cd596a148239/F4.jpg", "media/a0893598cd596a148239/F5.jpg", "media/a0893598cd596a148239/F6.jpg", "media/a0893598cd596a148239/F7.jpg", "media/a0893598cd596a148239/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white and purple SBS Transit double-decker bus) or B (silver hatchback car)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f0ddd60f38a05ff4b0767ea1a949c80dcfe50c2a398a5baec0d15381bf07143a"}
{"qa_id": "MOSE__31ac3d98__object_world_motion_direction__00", "video_id": "MOSE__31ac3d98", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (large grey elephant walking away) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/2bc6b57336f4c3040ff6/F0.jpg", "media/2bc6b57336f4c3040ff6/F1.jpg", "media/2bc6b57336f4c3040ff6/F2.jpg", "media/2bc6b57336f4c3040ff6/F3.jpg", "media/2bc6b57336f4c3040ff6/F4.jpg", "media/2bc6b57336f4c3040ff6/F5.jpg", "media/2bc6b57336f4c3040ff6/F6.jpg", "media/2bc6b57336f4c3040ff6/F7.jpg", "media/2bc6b57336f4c3040ff6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (large grey elephant walking away) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3d196ee139042acc9a0675ce9a74828b181d952532076a9f56ab7bc059f8064"}
{"qa_id": "MOSE__31ac3d98__camera_object_distance_change__01", "video_id": "MOSE__31ac3d98", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object A (large grey elephant walking away) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2bc6b57336f4c3040ff6/F0.jpg", "media/2bc6b57336f4c3040ff6/F1.jpg", "media/2bc6b57336f4c3040ff6/F2.jpg", "media/2bc6b57336f4c3040ff6/F3.jpg", "media/2bc6b57336f4c3040ff6/F4.jpg", "media/2bc6b57336f4c3040ff6/F5.jpg", "media/2bc6b57336f4c3040ff6/F6.jpg", "media/2bc6b57336f4c3040ff6/F7.jpg", "media/2bc6b57336f4c3040ff6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (large grey elephant walking away) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8d735f4f61c7b8a4caf9252d219b686e668e443484de9638a1cb3c189a9d895"}
{"qa_id": "MOSE__31b09833__camera_translation_direction__00", "video_id": "MOSE__31b09833", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "right", "forward"], "frame_paths": ["media/0231300d371486f39fa8/F0.jpg", "media/0231300d371486f39fa8/F1.jpg", "media/0231300d371486f39fa8/F2.jpg", "media/0231300d371486f39fa8/F3.jpg", "media/0231300d371486f39fa8/F4.jpg", "media/0231300d371486f39fa8/F5.jpg", "media/0231300d371486f39fa8/F6.jpg", "media/0231300d371486f39fa8/F7.jpg", "media/0231300d371486f39fa8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84415eff3289c6d6aa6ed62bfa1a28ca5e1e87c3854f4db4f0678f2500b1cce9"}
{"qa_id": "MOSE__31b09833__camera_view_direction_change__01", "video_id": "MOSE__31b09833", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/0231300d371486f39fa8/F0.jpg", "media/0231300d371486f39fa8/F1.jpg", "media/0231300d371486f39fa8/F2.jpg", "media/0231300d371486f39fa8/F3.jpg", "media/0231300d371486f39fa8/F4.jpg", "media/0231300d371486f39fa8/F5.jpg", "media/0231300d371486f39fa8/F6.jpg", "media/0231300d371486f39fa8/F7.jpg", "media/0231300d371486f39fa8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa5bbe0f773427eec251354ffb864b848dea2a22cd4db093baa7aa2643832758"}
{"qa_id": "MOSE__31b09833__camera_object_distance_change__02", "video_id": "MOSE__31b09833", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (white horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/7e023084172977f0f93f/F0.jpg", "media/7e023084172977f0f93f/F1.jpg", "media/7e023084172977f0f93f/F2.jpg", "media/7e023084172977f0f93f/F3.jpg", "media/7e023084172977f0f93f/F4.jpg", "media/7e023084172977f0f93f/F5.jpg", "media/7e023084172977f0f93f/F6.jpg", "media/7e023084172977f0f93f/F7.jpg", "media/7e023084172977f0f93f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (white horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ccda324e811bf8271cdb213c2dc04f401574f8db25df574b9bfd823ba08c1a3c"}
{"qa_id": "MOSE__31b09833__object_pair_distance_change__03", "video_id": "MOSE__31b09833", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did the 3D distance between object A (white horse) and object B (brown horse on the far right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/1328c54e88ca0e39246e/F0.jpg", "media/1328c54e88ca0e39246e/F1.jpg", "media/1328c54e88ca0e39246e/F2.jpg", "media/1328c54e88ca0e39246e/F3.jpg", "media/1328c54e88ca0e39246e/F4.jpg", "media/1328c54e88ca0e39246e/F5.jpg", "media/1328c54e88ca0e39246e/F6.jpg", "media/1328c54e88ca0e39246e/F7.jpg", "media/1328c54e88ca0e39246e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (white horse) and object B (brown horse on the far right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1b008bd8f903d1ed0736da256fb824559ac8ab0cd8478150f9d42633feaa8d5"}
{"qa_id": "MOSE__31b09833__object_travel_distance_comparison__04", "video_id": "MOSE__31b09833", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white horse) or B (brown horse on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1328c54e88ca0e39246e/F0.jpg", "media/1328c54e88ca0e39246e/F1.jpg", "media/1328c54e88ca0e39246e/F2.jpg", "media/1328c54e88ca0e39246e/F3.jpg", "media/1328c54e88ca0e39246e/F4.jpg", "media/1328c54e88ca0e39246e/F5.jpg", "media/1328c54e88ca0e39246e/F6.jpg", "media/1328c54e88ca0e39246e/F7.jpg", "media/1328c54e88ca0e39246e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white horse) or B (brown horse on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e6d320b2b6e8bed690b49ebe9b9b5908ec2062b71ae29b7d208972d2f11a3961"}
{"qa_id": "MOSE__32e5d721__object_world_motion_direction__00", "video_id": "MOSE__32e5d721", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (large elephant in the foreground) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "up", "forward", "down"], "frame_paths": ["media/8401decd34d66c51ae40/F0.jpg", "media/8401decd34d66c51ae40/F1.jpg", "media/8401decd34d66c51ae40/F2.jpg", "media/8401decd34d66c51ae40/F3.jpg", "media/8401decd34d66c51ae40/F4.jpg", "media/8401decd34d66c51ae40/F5.jpg", "media/8401decd34d66c51ae40/F6.jpg", "media/8401decd34d66c51ae40/F7.jpg", "media/8401decd34d66c51ae40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (large elephant in the foreground) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fdb1f96f3caa7ee5c548a65bb1534c59fc5deecc3c84c13de4bd2cd24b08a308"}
{"qa_id": "MOSE__32ecef67__object_pair_distance_change__00", "video_id": "MOSE__32ecef67", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did the 3D distance between object A (tiger standing on rocks on the left side) and object B (tiger walking on the right side) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/93578b92ec7601677b24/F0.jpg", "media/93578b92ec7601677b24/F1.jpg", "media/93578b92ec7601677b24/F2.jpg", "media/93578b92ec7601677b24/F3.jpg", "media/93578b92ec7601677b24/F4.jpg", "media/93578b92ec7601677b24/F5.jpg", "media/93578b92ec7601677b24/F6.jpg", "media/93578b92ec7601677b24/F7.jpg", "media/93578b92ec7601677b24/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (tiger standing on rocks on the left side) and object B (tiger walking on the right side) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62693120ac13134c2ecc06926ceba81d93c38a76e68845238fac625da5ffd78a"}
{"qa_id": "MOSE__33015509__camera_view_direction_change__00", "video_id": "MOSE__33015509", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/c5ab10955e30d3edfe35/F0.jpg", "media/c5ab10955e30d3edfe35/F1.jpg", "media/c5ab10955e30d3edfe35/F2.jpg", "media/c5ab10955e30d3edfe35/F3.jpg", "media/c5ab10955e30d3edfe35/F4.jpg", "media/c5ab10955e30d3edfe35/F5.jpg", "media/c5ab10955e30d3edfe35/F6.jpg", "media/c5ab10955e30d3edfe35/F7.jpg", "media/c5ab10955e30d3edfe35/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "MOSE__33015509__object_pair_distance_change__01", "video_id": "MOSE__33015509", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did the 3D distance between object A (large dark brown horse) and object B (medium brown horse) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/86fe6e51a4b09daf9f5b/F0.jpg", "media/86fe6e51a4b09daf9f5b/F1.jpg", "media/86fe6e51a4b09daf9f5b/F2.jpg", "media/86fe6e51a4b09daf9f5b/F3.jpg", "media/86fe6e51a4b09daf9f5b/F4.jpg", "media/86fe6e51a4b09daf9f5b/F5.jpg", "media/86fe6e51a4b09daf9f5b/F6.jpg", "media/86fe6e51a4b09daf9f5b/F7.jpg", "media/86fe6e51a4b09daf9f5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (large dark brown horse) and object B (medium brown horse) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "70f8ad5359b37e3de493f0226df3f5ade0cf998004e82208d7e4920aa9e6e561"}
{"qa_id": "MOSE__33015509__object_travel_distance_comparison__02", "video_id": "MOSE__33015509", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large dark brown horse) or B (medium brown horse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/86fe6e51a4b09daf9f5b/F0.jpg", "media/86fe6e51a4b09daf9f5b/F1.jpg", "media/86fe6e51a4b09daf9f5b/F2.jpg", "media/86fe6e51a4b09daf9f5b/F3.jpg", "media/86fe6e51a4b09daf9f5b/F4.jpg", "media/86fe6e51a4b09daf9f5b/F5.jpg", "media/86fe6e51a4b09daf9f5b/F6.jpg", "media/86fe6e51a4b09daf9f5b/F7.jpg", "media/86fe6e51a4b09daf9f5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large dark brown horse) or B (medium brown horse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f5dd3fadcb5d1984de81fa9d5e00b0f69815c17e770f03dd58d74839902bb56"}
{"qa_id": "MOSE__33602f6b__object_pair_distance_change__00", "video_id": "MOSE__33602f6b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did the 3D distance between object A (young tiger cub standing on a large rock) and object B (young tiger cub near the enclosure barrier) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/aa26b0eb1e2b69c85560/F0.jpg", "media/aa26b0eb1e2b69c85560/F1.jpg", "media/aa26b0eb1e2b69c85560/F2.jpg", "media/aa26b0eb1e2b69c85560/F3.jpg", "media/aa26b0eb1e2b69c85560/F4.jpg", "media/aa26b0eb1e2b69c85560/F5.jpg", "media/aa26b0eb1e2b69c85560/F6.jpg", "media/aa26b0eb1e2b69c85560/F7.jpg", "media/aa26b0eb1e2b69c85560/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (young tiger cub standing on a large rock) and object B (young tiger cub near the enclosure barrier) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "04cef98cdd4a94a4dd4dfa8e1fca96ed4929c806f38332791e410ac4c85676f1"}
{"qa_id": "MOSE__33602f6b__object_travel_distance_comparison__01", "video_id": "MOSE__33602f6b", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (young tiger cub standing on a large rock) or B (young tiger cub near the enclosure barrier)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/aa26b0eb1e2b69c85560/F0.jpg", "media/aa26b0eb1e2b69c85560/F1.jpg", "media/aa26b0eb1e2b69c85560/F2.jpg", "media/aa26b0eb1e2b69c85560/F3.jpg", "media/aa26b0eb1e2b69c85560/F4.jpg", "media/aa26b0eb1e2b69c85560/F5.jpg", "media/aa26b0eb1e2b69c85560/F6.jpg", "media/aa26b0eb1e2b69c85560/F7.jpg", "media/aa26b0eb1e2b69c85560/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (young tiger cub standing on a large rock) or B (young tiger cub near the enclosure barrier)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eeab2f346b204860d4a22517521d56d7cb52cc73a4c38c9c013f04ae4ac9b76e"}
{"qa_id": "MOSE__34a5f964__object_world_motion_direction__00", "video_id": "MOSE__34a5f964", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (person wearing a blue and white patterned shirt and dark pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "right", "left"], "frame_paths": ["media/d6bc05852c0612efc94a/F0.jpg", "media/d6bc05852c0612efc94a/F1.jpg", "media/d6bc05852c0612efc94a/F2.jpg", "media/d6bc05852c0612efc94a/F3.jpg", "media/d6bc05852c0612efc94a/F4.jpg", "media/d6bc05852c0612efc94a/F5.jpg", "media/d6bc05852c0612efc94a/F6.jpg", "media/d6bc05852c0612efc94a/F7.jpg", "media/d6bc05852c0612efc94a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (person wearing a blue and white patterned shirt and dark pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "79e62e75824faa9a61a1336e48dc91d14663457bd978272960f254f0d7cd2bef"}
{"qa_id": "MOSE__34a5f964__camera_object_distance_change__01", "video_id": "MOSE__34a5f964", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object B (person wearing a blue and white patterned shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d6bc05852c0612efc94a/F0.jpg", "media/d6bc05852c0612efc94a/F1.jpg", "media/d6bc05852c0612efc94a/F2.jpg", "media/d6bc05852c0612efc94a/F3.jpg", "media/d6bc05852c0612efc94a/F4.jpg", "media/d6bc05852c0612efc94a/F5.jpg", "media/d6bc05852c0612efc94a/F6.jpg", "media/d6bc05852c0612efc94a/F7.jpg", "media/d6bc05852c0612efc94a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (person wearing a blue and white patterned shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f292baa34b0d164cc18d5b2ce350cb230fd2e5a00d7073dd2aa53c748168e5bc"}
{"qa_id": "MOSE__34a5f964__object_pair_distance_change__02", "video_id": "MOSE__34a5f964", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (black motorcycle with a blue bag hanging on the handlebars) and object B (person wearing a blue and white patterned shirt and dark pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/24efd23465dadc33a8f4/F0.jpg", "media/24efd23465dadc33a8f4/F1.jpg", "media/24efd23465dadc33a8f4/F2.jpg", "media/24efd23465dadc33a8f4/F3.jpg", "media/24efd23465dadc33a8f4/F4.jpg", "media/24efd23465dadc33a8f4/F5.jpg", "media/24efd23465dadc33a8f4/F6.jpg", "media/24efd23465dadc33a8f4/F7.jpg", "media/24efd23465dadc33a8f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (black motorcycle with a blue bag hanging on the handlebars) and object B (person wearing a blue and white patterned shirt and dark pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f77568308a87a99229a9e22d3a9d2e8d72240a9648233054bf825c60e791bb41"}
{"qa_id": "MOSE__34a5f964__object_travel_distance_comparison__03", "video_id": "MOSE__34a5f964", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black motorcycle with a blue bag hanging on the handlebars) or B (person wearing a blue and white patterned shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/24efd23465dadc33a8f4/F0.jpg", "media/24efd23465dadc33a8f4/F1.jpg", "media/24efd23465dadc33a8f4/F2.jpg", "media/24efd23465dadc33a8f4/F3.jpg", "media/24efd23465dadc33a8f4/F4.jpg", "media/24efd23465dadc33a8f4/F5.jpg", "media/24efd23465dadc33a8f4/F6.jpg", "media/24efd23465dadc33a8f4/F7.jpg", "media/24efd23465dadc33a8f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black motorcycle with a blue bag hanging on the handlebars) or B (person wearing a blue and white patterned shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66d97b3702da33a90abf2fce37d33921e980eb83ffc63965ae190d37f3ab3c68"}
{"qa_id": "MOSE__34ccea00__object_world_motion_direction__00", "video_id": "MOSE__34ccea00", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (light brown bear cub) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "backward", "up", "forward"], "frame_paths": ["media/af85f4015b06c04cd855/F0.jpg", "media/af85f4015b06c04cd855/F1.jpg", "media/af85f4015b06c04cd855/F2.jpg", "media/af85f4015b06c04cd855/F3.jpg", "media/af85f4015b06c04cd855/F4.jpg", "media/af85f4015b06c04cd855/F5.jpg", "media/af85f4015b06c04cd855/F6.jpg", "media/af85f4015b06c04cd855/F7.jpg", "media/af85f4015b06c04cd855/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (light brown bear cub) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5562c18b3243177b343f26e1b85b7319d23695425a2f43a2a0da6ad96665ea1"}
{"qa_id": "MOSE__34ccea00__camera_object_distance_change__01", "video_id": "MOSE__34ccea00", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (light brown bear cub) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/af85f4015b06c04cd855/F0.jpg", "media/af85f4015b06c04cd855/F1.jpg", "media/af85f4015b06c04cd855/F2.jpg", "media/af85f4015b06c04cd855/F3.jpg", "media/af85f4015b06c04cd855/F4.jpg", "media/af85f4015b06c04cd855/F5.jpg", "media/af85f4015b06c04cd855/F6.jpg", "media/af85f4015b06c04cd855/F7.jpg", "media/af85f4015b06c04cd855/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (light brown bear cub) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f584f4ce67cd1970ede839ced844da98f3cac61d43a79c1adc78ec8dbd1edcc6"}
{"qa_id": "MOSE__34ccea00__object_pair_distance_change__02", "video_id": "MOSE__34ccea00", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (light brown bear cub) and object B (dark brown bear cub) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/51825303e3c946bbf0c7/F0.jpg", "media/51825303e3c946bbf0c7/F1.jpg", "media/51825303e3c946bbf0c7/F2.jpg", "media/51825303e3c946bbf0c7/F3.jpg", "media/51825303e3c946bbf0c7/F4.jpg", "media/51825303e3c946bbf0c7/F5.jpg", "media/51825303e3c946bbf0c7/F6.jpg", "media/51825303e3c946bbf0c7/F7.jpg", "media/51825303e3c946bbf0c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (light brown bear cub) and object B (dark brown bear cub) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "882987b71975cfca66e2d67ab9eab72439e67170d027a43200d004e0076dcf06"}
{"qa_id": "MOSE__34ccea00__object_travel_distance_comparison__03", "video_id": "MOSE__34ccea00", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown bear cub) or B (dark brown bear cub)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/51825303e3c946bbf0c7/F0.jpg", "media/51825303e3c946bbf0c7/F1.jpg", "media/51825303e3c946bbf0c7/F2.jpg", "media/51825303e3c946bbf0c7/F3.jpg", "media/51825303e3c946bbf0c7/F4.jpg", "media/51825303e3c946bbf0c7/F5.jpg", "media/51825303e3c946bbf0c7/F6.jpg", "media/51825303e3c946bbf0c7/F7.jpg", "media/51825303e3c946bbf0c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown bear cub) or B (dark brown bear cub)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "409a08aa3cacd72587122f81ba87938630109c567d662d9abb13b81ec4806252"}
{"qa_id": "MOSE__3592425a__camera_view_direction_change__00", "video_id": "MOSE__3592425a", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/f359397d48ce4844cac4/F0.jpg", "media/f359397d48ce4844cac4/F1.jpg", "media/f359397d48ce4844cac4/F2.jpg", "media/f359397d48ce4844cac4/F3.jpg", "media/f359397d48ce4844cac4/F4.jpg", "media/f359397d48ce4844cac4/F5.jpg", "media/f359397d48ce4844cac4/F6.jpg", "media/f359397d48ce4844cac4/F7.jpg", "media/f359397d48ce4844cac4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13dac79e2013fbcd040093fc210e0a05f869f71c449dfee4a4d2a9a58dd4088f"}
{"qa_id": "MOSE__3592425a__object_world_motion_direction__01", "video_id": "MOSE__3592425a", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (large flat fish with a mottled grey and white pattern resting on the rocks) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "down", "forward", "backward", "left", "right"], "frame_paths": ["media/0253d53ad3854b1984b4/F0.jpg", "media/0253d53ad3854b1984b4/F1.jpg", "media/0253d53ad3854b1984b4/F2.jpg", "media/0253d53ad3854b1984b4/F3.jpg", "media/0253d53ad3854b1984b4/F4.jpg", "media/0253d53ad3854b1984b4/F5.jpg", "media/0253d53ad3854b1984b4/F6.jpg", "media/0253d53ad3854b1984b4/F7.jpg", "media/0253d53ad3854b1984b4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (large flat fish with a mottled grey and white pattern resting on the rocks) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. forward\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca0f936c5517274646a6b102c3450c796562c1ed27cbe3b5dbb112dd1ba62f06"}
{"qa_id": "MOSE__37549a79__object_world_motion_direction__00", "video_id": "MOSE__37549a79", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (black bull with white markings on its back) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "left", "down", "right"], "frame_paths": ["media/31eb861d385eba9cba9e/F0.jpg", "media/31eb861d385eba9cba9e/F1.jpg", "media/31eb861d385eba9cba9e/F2.jpg", "media/31eb861d385eba9cba9e/F3.jpg", "media/31eb861d385eba9cba9e/F4.jpg", "media/31eb861d385eba9cba9e/F5.jpg", "media/31eb861d385eba9cba9e/F6.jpg", "media/31eb861d385eba9cba9e/F7.jpg", "media/31eb861d385eba9cba9e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (black bull with white markings on its back) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5772092d68cae49d38f3ce3209038b0032d58d547533d05cd85717d84e1c8a25"}
{"qa_id": "MOSE__376dd830__camera_translation_direction__00", "video_id": "MOSE__376dd830", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "backward", "up", "down"], "frame_paths": ["media/dc2da7f591d2da7bf7a0/F0.jpg", "media/dc2da7f591d2da7bf7a0/F1.jpg", "media/dc2da7f591d2da7bf7a0/F2.jpg", "media/dc2da7f591d2da7bf7a0/F3.jpg", "media/dc2da7f591d2da7bf7a0/F4.jpg", "media/dc2da7f591d2da7bf7a0/F5.jpg", "media/dc2da7f591d2da7bf7a0/F6.jpg", "media/dc2da7f591d2da7bf7a0/F7.jpg", "media/dc2da7f591d2da7bf7a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "575ae1aea6e86a826538d80fb160d0a4d233eaa3628c69664c04f9d7b46bafe4"}
{"qa_id": "MOSE__37af9ac1__object_world_motion_direction__00", "video_id": "MOSE__37af9ac1", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object A (black and white chicken with a red comb) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "left", "up", "forward"], "frame_paths": ["media/ac270045f197e9d00ae6/F0.jpg", "media/ac270045f197e9d00ae6/F1.jpg", "media/ac270045f197e9d00ae6/F2.jpg", "media/ac270045f197e9d00ae6/F3.jpg", "media/ac270045f197e9d00ae6/F4.jpg", "media/ac270045f197e9d00ae6/F5.jpg", "media/ac270045f197e9d00ae6/F6.jpg", "media/ac270045f197e9d00ae6/F7.jpg", "media/ac270045f197e9d00ae6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object A (black and white chicken with a red comb) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb213ad80083eea349f22e28a99b8fe57c640a99e4842bbec1784a93eae009c4"}
{"qa_id": "MOSE__37af9ac1__camera_object_distance_change__01", "video_id": "MOSE__37af9ac1", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object A (black and white chicken with a red comb) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ac270045f197e9d00ae6/F0.jpg", "media/ac270045f197e9d00ae6/F1.jpg", "media/ac270045f197e9d00ae6/F2.jpg", "media/ac270045f197e9d00ae6/F3.jpg", "media/ac270045f197e9d00ae6/F4.jpg", "media/ac270045f197e9d00ae6/F5.jpg", "media/ac270045f197e9d00ae6/F6.jpg", "media/ac270045f197e9d00ae6/F7.jpg", "media/ac270045f197e9d00ae6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (black and white chicken with a red comb) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5b825805d6913d4e6548ba4a642b805ea6a436a1c00b5771402abec8af9bca5"}
{"qa_id": "MOSE__37af9ac1__object_pair_distance_change__02", "video_id": "MOSE__37af9ac1", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did the 3D distance between object A (black and white chicken with a red comb) and object B (grey and white goose with a long neck) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/9f1425983affa87d7d93/F0.jpg", "media/9f1425983affa87d7d93/F1.jpg", "media/9f1425983affa87d7d93/F2.jpg", "media/9f1425983affa87d7d93/F3.jpg", "media/9f1425983affa87d7d93/F4.jpg", "media/9f1425983affa87d7d93/F5.jpg", "media/9f1425983affa87d7d93/F6.jpg", "media/9f1425983affa87d7d93/F7.jpg", "media/9f1425983affa87d7d93/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (black and white chicken with a red comb) and object B (grey and white goose with a long neck) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "63fc2a07ca79aa83addee51d9cf8cb81757674e73393eb9443bd6f3a97835883"}
{"qa_id": "MOSE__37af9ac1__object_travel_distance_comparison__03", "video_id": "MOSE__37af9ac1", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black and white chicken with a red comb) or B (grey and white goose with a long neck)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/9f1425983affa87d7d93/F0.jpg", "media/9f1425983affa87d7d93/F1.jpg", "media/9f1425983affa87d7d93/F2.jpg", "media/9f1425983affa87d7d93/F3.jpg", "media/9f1425983affa87d7d93/F4.jpg", "media/9f1425983affa87d7d93/F5.jpg", "media/9f1425983affa87d7d93/F6.jpg", "media/9f1425983affa87d7d93/F7.jpg", "media/9f1425983affa87d7d93/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black and white chicken with a red comb) or B (grey and white goose with a long neck)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff549061555a4283254bf77455f40d5ef6742f1290e327dc585bb8ba3f38688d"}
{"qa_id": "MOSE__37b07a28__camera_view_direction_change__00", "video_id": "MOSE__37b07a28", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/bc2dbd336a5918e54f35/F0.jpg", "media/bc2dbd336a5918e54f35/F1.jpg", "media/bc2dbd336a5918e54f35/F2.jpg", "media/bc2dbd336a5918e54f35/F3.jpg", "media/bc2dbd336a5918e54f35/F4.jpg", "media/bc2dbd336a5918e54f35/F5.jpg", "media/bc2dbd336a5918e54f35/F6.jpg", "media/bc2dbd336a5918e54f35/F7.jpg", "media/bc2dbd336a5918e54f35/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc01c8f565adc7d70c3dfe8415bb1beb47a7031980d38a04b07a94aa6f06829f"}
{"qa_id": "MOSE__37b07a28__camera_object_distance_change__01", "video_id": "MOSE__37b07a28", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did object B (large grey elephant walking on grass) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/87317e627d371e47ac52/F0.jpg", "media/87317e627d371e47ac52/F1.jpg", "media/87317e627d371e47ac52/F2.jpg", "media/87317e627d371e47ac52/F3.jpg", "media/87317e627d371e47ac52/F4.jpg", "media/87317e627d371e47ac52/F5.jpg", "media/87317e627d371e47ac52/F6.jpg", "media/87317e627d371e47ac52/F7.jpg", "media/87317e627d371e47ac52/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object B (large grey elephant walking on grass) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7950d16f0a2e2de0d86d6e7f32e478e9bb1dd55ab16fb64469f76bc84fa51cc"}
{"qa_id": "MOSE__37b07a28__object_pair_distance_change__02", "video_id": "MOSE__37b07a28", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (large grey elephant walking on grass) and object B (large grey elephant walking on grass) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b28e4cd97151dd9fd61d/F0.jpg", "media/b28e4cd97151dd9fd61d/F1.jpg", "media/b28e4cd97151dd9fd61d/F2.jpg", "media/b28e4cd97151dd9fd61d/F3.jpg", "media/b28e4cd97151dd9fd61d/F4.jpg", "media/b28e4cd97151dd9fd61d/F5.jpg", "media/b28e4cd97151dd9fd61d/F6.jpg", "media/b28e4cd97151dd9fd61d/F7.jpg", "media/b28e4cd97151dd9fd61d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (large grey elephant walking on grass) and object B (large grey elephant walking on grass) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9e4fdccfc0bd97b166f3d752de3700309d62ca81489e7d56eb20b10cb6a6dfa5"}
{"qa_id": "MOSE__37b07a28__object_travel_distance_comparison__03", "video_id": "MOSE__37b07a28", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant walking on grass) or B (large grey elephant walking on grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/b28e4cd97151dd9fd61d/F0.jpg", "media/b28e4cd97151dd9fd61d/F1.jpg", "media/b28e4cd97151dd9fd61d/F2.jpg", "media/b28e4cd97151dd9fd61d/F3.jpg", "media/b28e4cd97151dd9fd61d/F4.jpg", "media/b28e4cd97151dd9fd61d/F5.jpg", "media/b28e4cd97151dd9fd61d/F6.jpg", "media/b28e4cd97151dd9fd61d/F7.jpg", "media/b28e4cd97151dd9fd61d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant walking on grass) or B (large grey elephant walking on grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1118a0f95c39236c01b08988c7e3dff4d5bbfd3220246a66f89de93982f40e7"}
{"qa_id": "MOSE__37bf7edf__object_world_motion_direction__00", "video_id": "MOSE__37bf7edf", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (large grey elephant walking away from the camera) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "down", "up", "left"], "frame_paths": ["media/fa4b6bf2cd700776feee/F0.jpg", "media/fa4b6bf2cd700776feee/F1.jpg", "media/fa4b6bf2cd700776feee/F2.jpg", "media/fa4b6bf2cd700776feee/F3.jpg", "media/fa4b6bf2cd700776feee/F4.jpg", "media/fa4b6bf2cd700776feee/F5.jpg", "media/fa4b6bf2cd700776feee/F6.jpg", "media/fa4b6bf2cd700776feee/F7.jpg", "media/fa4b6bf2cd700776feee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (large grey elephant walking away from the camera) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "57dcc68bf7ec903f04afc8c0e229410e6fb4a084a7ec51e12c46d7be861efd7a"}
{"qa_id": "MOSE__37bf7edf__camera_object_distance_change__01", "video_id": "MOSE__37bf7edf", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object B (elephant in the background near the trees) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5bc54af09d048ff8d45b/F0.jpg", "media/5bc54af09d048ff8d45b/F1.jpg", "media/5bc54af09d048ff8d45b/F2.jpg", "media/5bc54af09d048ff8d45b/F3.jpg", "media/5bc54af09d048ff8d45b/F4.jpg", "media/5bc54af09d048ff8d45b/F5.jpg", "media/5bc54af09d048ff8d45b/F6.jpg", "media/5bc54af09d048ff8d45b/F7.jpg", "media/5bc54af09d048ff8d45b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (elephant in the background near the trees) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ae295d582513bd6b1db511f4ae4ed8aeb31b0a9c536a45241158a87c9a7c8d2e"}
{"qa_id": "MOSE__37bf7edf__object_pair_distance_change__02", "video_id": "MOSE__37bf7edf", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (large grey elephant walking away from the camera) and object B (elephant in the background near the trees) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/63400f14cf92577b2d6c/F0.jpg", "media/63400f14cf92577b2d6c/F1.jpg", "media/63400f14cf92577b2d6c/F2.jpg", "media/63400f14cf92577b2d6c/F3.jpg", "media/63400f14cf92577b2d6c/F4.jpg", "media/63400f14cf92577b2d6c/F5.jpg", "media/63400f14cf92577b2d6c/F6.jpg", "media/63400f14cf92577b2d6c/F7.jpg", "media/63400f14cf92577b2d6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (large grey elephant walking away from the camera) and object B (elephant in the background near the trees) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8aed7161a0fcc82ac68281b2e2b3c54c14970d2c0c7f0a56f9c45cef5991f62"}
{"qa_id": "MOSE__37bf7edf__object_travel_distance_comparison__03", "video_id": "MOSE__37bf7edf", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant walking away from the camera) or B (elephant in the background near the trees)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/63400f14cf92577b2d6c/F0.jpg", "media/63400f14cf92577b2d6c/F1.jpg", "media/63400f14cf92577b2d6c/F2.jpg", "media/63400f14cf92577b2d6c/F3.jpg", "media/63400f14cf92577b2d6c/F4.jpg", "media/63400f14cf92577b2d6c/F5.jpg", "media/63400f14cf92577b2d6c/F6.jpg", "media/63400f14cf92577b2d6c/F7.jpg", "media/63400f14cf92577b2d6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant walking away from the camera) or B (elephant in the background near the trees)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8230b919a9803244a87ee2a33ea259caf9973461cd3f3467c7db29a05c7990b2"}
{"qa_id": "MOSE__395224b3__camera_view_direction_change__00", "video_id": "MOSE__395224b3", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/8893b6c0add4eb03eb3d/F0.jpg", "media/8893b6c0add4eb03eb3d/F1.jpg", "media/8893b6c0add4eb03eb3d/F2.jpg", "media/8893b6c0add4eb03eb3d/F3.jpg", "media/8893b6c0add4eb03eb3d/F4.jpg", "media/8893b6c0add4eb03eb3d/F5.jpg", "media/8893b6c0add4eb03eb3d/F6.jpg", "media/8893b6c0add4eb03eb3d/F7.jpg", "media/8893b6c0add4eb03eb3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "MOSE__39cc637e__camera_view_direction_change__00", "video_id": "MOSE__39cc637e", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/210137f6bfb3c3068cc0/F0.jpg", "media/210137f6bfb3c3068cc0/F1.jpg", "media/210137f6bfb3c3068cc0/F2.jpg", "media/210137f6bfb3c3068cc0/F3.jpg", "media/210137f6bfb3c3068cc0/F4.jpg", "media/210137f6bfb3c3068cc0/F5.jpg", "media/210137f6bfb3c3068cc0/F6.jpg", "media/210137f6bfb3c3068cc0/F7.jpg", "media/210137f6bfb3c3068cc0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c62e4cce265ee0250831b207eb7336b04a7d9ae9f67b14a17d5a22e813943899"}
{"qa_id": "MOSE__3aab2d13__camera_translation_direction__00", "video_id": "MOSE__3aab2d13", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "backward", "left", "up"], "frame_paths": ["media/fd3c56cbb23129b769c3/F0.jpg", "media/fd3c56cbb23129b769c3/F1.jpg", "media/fd3c56cbb23129b769c3/F2.jpg", "media/fd3c56cbb23129b769c3/F3.jpg", "media/fd3c56cbb23129b769c3/F4.jpg", "media/fd3c56cbb23129b769c3/F5.jpg", "media/fd3c56cbb23129b769c3/F6.jpg", "media/fd3c56cbb23129b769c3/F7.jpg", "media/fd3c56cbb23129b769c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a9a91b5a7c2d64b922f62c0cd8919717468bafe2143aa83c3f039c8c4d893cba"}
{"qa_id": "MOSE__3aab2d13__object_world_motion_direction__01", "video_id": "MOSE__3aab2d13", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object B (smaller turtle with a smooth brown shell) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "backward", "left", "up"], "frame_paths": ["media/28f16b86c3e81231dce5/F0.jpg", "media/28f16b86c3e81231dce5/F1.jpg", "media/28f16b86c3e81231dce5/F2.jpg", "media/28f16b86c3e81231dce5/F3.jpg", "media/28f16b86c3e81231dce5/F4.jpg", "media/28f16b86c3e81231dce5/F5.jpg", "media/28f16b86c3e81231dce5/F6.jpg", "media/28f16b86c3e81231dce5/F7.jpg", "media/28f16b86c3e81231dce5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object B (smaller turtle with a smooth brown shell) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31284b7da9d46a458cbed972422a4ecdb2018716b03d6321e15ffe2885af69e4"}
{"qa_id": "MOSE__3aab2d13__object_travel_distance_comparison__02", "video_id": "MOSE__3aab2d13", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (turtle with a patterned brown shell) or B (smaller turtle with a smooth brown shell)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/e00d3e8ea1989077cf49/F0.jpg", "media/e00d3e8ea1989077cf49/F1.jpg", "media/e00d3e8ea1989077cf49/F2.jpg", "media/e00d3e8ea1989077cf49/F3.jpg", "media/e00d3e8ea1989077cf49/F4.jpg", "media/e00d3e8ea1989077cf49/F5.jpg", "media/e00d3e8ea1989077cf49/F6.jpg", "media/e00d3e8ea1989077cf49/F7.jpg", "media/e00d3e8ea1989077cf49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (turtle with a patterned brown shell) or B (smaller turtle with a smooth brown shell)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d97fd3fe8e2e51444f5fbf9807c8ab8ca59a587358764bb0205a93d80fb1b3a0"}
{"qa_id": "MOSE__3afd02d7__object_pair_distance_change__00", "video_id": "MOSE__3afd02d7", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (motocross rider in blue and white gear on a white dirt bike) and object B (motocross rider in black and yellow gear on a white dirt bike) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ae8cbe41fffdb758733d/F0.jpg", "media/ae8cbe41fffdb758733d/F1.jpg", "media/ae8cbe41fffdb758733d/F2.jpg", "media/ae8cbe41fffdb758733d/F3.jpg", "media/ae8cbe41fffdb758733d/F4.jpg", "media/ae8cbe41fffdb758733d/F5.jpg", "media/ae8cbe41fffdb758733d/F6.jpg", "media/ae8cbe41fffdb758733d/F7.jpg", "media/ae8cbe41fffdb758733d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (motocross rider in blue and white gear on a white dirt bike) and object B (motocross rider in black and yellow gear on a white dirt bike) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a14a317c9404ee4b332fd3ce6daab3013816ed552e6f5534ae7865dcaacf4cad"}
{"qa_id": "MOSE__3bd7facd__object_world_motion_direction__00", "video_id": "MOSE__3bd7facd", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (Kunming Airlines airplane with a white fuselage and a blue tail featuring a gold logo) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "down", "right", "up", "backward"], "frame_paths": ["media/fd634cc38182c7b20acc/F0.jpg", "media/fd634cc38182c7b20acc/F1.jpg", "media/fd634cc38182c7b20acc/F2.jpg", "media/fd634cc38182c7b20acc/F3.jpg", "media/fd634cc38182c7b20acc/F4.jpg", "media/fd634cc38182c7b20acc/F5.jpg", "media/fd634cc38182c7b20acc/F6.jpg", "media/fd634cc38182c7b20acc/F7.jpg", "media/fd634cc38182c7b20acc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (Kunming Airlines airplane with a white fuselage and a blue tail featuring a gold logo) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. down\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "316426edb1d99d600a1937a83979a6ba51974caeba9d9c62e8e13eba8ae1e999"}
{"qa_id": "MOSE__3d08dadd__camera_view_direction_change__00", "video_id": "MOSE__3d08dadd", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/f3a5d45148398b7f2286/F0.jpg", "media/f3a5d45148398b7f2286/F1.jpg", "media/f3a5d45148398b7f2286/F2.jpg", "media/f3a5d45148398b7f2286/F3.jpg", "media/f3a5d45148398b7f2286/F4.jpg", "media/f3a5d45148398b7f2286/F5.jpg", "media/f3a5d45148398b7f2286/F6.jpg", "media/f3a5d45148398b7f2286/F7.jpg", "media/f3a5d45148398b7f2286/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a2745d361a044b2569420928ab2ca17bcd7c31c4a41aabda3a0f9aaa0908851"}
{"qa_id": "MOSE__3d08dadd__object_world_motion_direction__01", "video_id": "MOSE__3d08dadd", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (light brown horse with a dark mane and tail) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "backward", "left", "right"], "frame_paths": ["media/197adefe8429368702e9/F0.jpg", "media/197adefe8429368702e9/F1.jpg", "media/197adefe8429368702e9/F2.jpg", "media/197adefe8429368702e9/F3.jpg", "media/197adefe8429368702e9/F4.jpg", "media/197adefe8429368702e9/F5.jpg", "media/197adefe8429368702e9/F6.jpg", "media/197adefe8429368702e9/F7.jpg", "media/197adefe8429368702e9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (light brown horse with a dark mane and tail) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "224947b5e540df18101a5aedda95803fcbdfe5e7a46db439dae4dc07970ebd12"}
{"qa_id": "MOSE__3d9a7023__object_world_motion_direction__00", "video_id": "MOSE__3d9a7023", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (yak with light brown and white fur) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "up", "forward", "down"], "frame_paths": ["media/b23fd99b7c95dcd51b91/F0.jpg", "media/b23fd99b7c95dcd51b91/F1.jpg", "media/b23fd99b7c95dcd51b91/F2.jpg", "media/b23fd99b7c95dcd51b91/F3.jpg", "media/b23fd99b7c95dcd51b91/F4.jpg", "media/b23fd99b7c95dcd51b91/F5.jpg", "media/b23fd99b7c95dcd51b91/F6.jpg", "media/b23fd99b7c95dcd51b91/F7.jpg", "media/b23fd99b7c95dcd51b91/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (yak with light brown and white fur) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bfbc2471dee29be9e4f8b548e89de717dd9f5aecb263147cb15f3e176236ee10"}
{"qa_id": "MOSE__3da63bd5__camera_translation_direction__00", "video_id": "MOSE__3da63bd5", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "forward", "backward"], "frame_paths": ["media/586679fc3732054b02b8/F0.jpg", "media/586679fc3732054b02b8/F1.jpg", "media/586679fc3732054b02b8/F2.jpg", "media/586679fc3732054b02b8/F3.jpg", "media/586679fc3732054b02b8/F4.jpg", "media/586679fc3732054b02b8/F5.jpg", "media/586679fc3732054b02b8/F6.jpg", "media/586679fc3732054b02b8/F7.jpg", "media/586679fc3732054b02b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "17a83212397b48b1fa34ec487271229113aa01ecd7d59096bb09869f69fc6b78"}
{"qa_id": "MOSE__3da63bd5__object_world_motion_direction__01", "video_id": "MOSE__3da63bd5", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (man in white t-shirt and blue jeans) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "right", "forward", "backward"], "frame_paths": ["media/97e5c452017a95b8ed46/F0.jpg", "media/97e5c452017a95b8ed46/F1.jpg", "media/97e5c452017a95b8ed46/F2.jpg", "media/97e5c452017a95b8ed46/F3.jpg", "media/97e5c452017a95b8ed46/F4.jpg", "media/97e5c452017a95b8ed46/F5.jpg", "media/97e5c452017a95b8ed46/F6.jpg", "media/97e5c452017a95b8ed46/F7.jpg", "media/97e5c452017a95b8ed46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (man in white t-shirt and blue jeans) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9f755c1769ed31fc8f40253cb9012ccbe702361b33220b08be236aa16d80100a"}
{"qa_id": "MOSE__3da63bd5__camera_object_distance_change__02", "video_id": "MOSE__3da63bd5", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object A (man in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/97e5c452017a95b8ed46/F0.jpg", "media/97e5c452017a95b8ed46/F1.jpg", "media/97e5c452017a95b8ed46/F2.jpg", "media/97e5c452017a95b8ed46/F3.jpg", "media/97e5c452017a95b8ed46/F4.jpg", "media/97e5c452017a95b8ed46/F5.jpg", "media/97e5c452017a95b8ed46/F6.jpg", "media/97e5c452017a95b8ed46/F7.jpg", "media/97e5c452017a95b8ed46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (man in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "30e8e9b27d3da8e4faf4e5f2674854c625efe0a417f6bd5a64f6f6263d6df3ab"}
{"qa_id": "MOSE__3e5ad020__camera_object_distance_change__00", "video_id": "MOSE__3e5ad020", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (black and white dog with a long tail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2c7e6c56c0cbc51aecf7/F0.jpg", "media/2c7e6c56c0cbc51aecf7/F1.jpg", "media/2c7e6c56c0cbc51aecf7/F2.jpg", "media/2c7e6c56c0cbc51aecf7/F3.jpg", "media/2c7e6c56c0cbc51aecf7/F4.jpg", "media/2c7e6c56c0cbc51aecf7/F5.jpg", "media/2c7e6c56c0cbc51aecf7/F6.jpg", "media/2c7e6c56c0cbc51aecf7/F7.jpg", "media/2c7e6c56c0cbc51aecf7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (black and white dog with a long tail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "091243469dbe4f3e48f7cb6fd0533e5fd921aead648ad2edaec544918ce660d4"}
{"qa_id": "MOSE__3f003746__object_world_motion_direction__00", "video_id": "MOSE__3f003746", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (motorcyclist in white and black racing suit with red and black helmet riding a yellow and black bike) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "right", "forward"], "frame_paths": ["media/557b2483a112940a58de/F0.jpg", "media/557b2483a112940a58de/F1.jpg", "media/557b2483a112940a58de/F2.jpg", "media/557b2483a112940a58de/F3.jpg", "media/557b2483a112940a58de/F4.jpg", "media/557b2483a112940a58de/F5.jpg", "media/557b2483a112940a58de/F6.jpg", "media/557b2483a112940a58de/F7.jpg", "media/557b2483a112940a58de/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g8", "B": "g9"}, "requested_object_ids": ["g8"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (motorcyclist in white and black racing suit with red and black helmet riding a yellow and black bike) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5f3f9a9754ef5f7a634394c3102536bee445c99099d861bf1cad9594bf8e868c"}
{"qa_id": "MOSE__3f313a2f__object_pair_distance_change__00", "video_id": "MOSE__3f313a2f", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did the 3D distance between object A (orange fish) and object B (pinkish fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8f76ef97d30d0106ac57/F0.jpg", "media/8f76ef97d30d0106ac57/F1.jpg", "media/8f76ef97d30d0106ac57/F2.jpg", "media/8f76ef97d30d0106ac57/F3.jpg", "media/8f76ef97d30d0106ac57/F4.jpg", "media/8f76ef97d30d0106ac57/F5.jpg", "media/8f76ef97d30d0106ac57/F6.jpg", "media/8f76ef97d30d0106ac57/F7.jpg", "media/8f76ef97d30d0106ac57/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did the 3D distance between object A (orange fish) and object B (pinkish fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b4ccc9a82a0c58ebbb6c9a5d58af73620be4168bb24b591aa6810912ac2a1a5"}
{"qa_id": "MOSE__4023fc77__camera_view_direction_change__00", "video_id": "MOSE__4023fc77", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/e695c0c4a4b4c23bcee2/F0.jpg", "media/e695c0c4a4b4c23bcee2/F1.jpg", "media/e695c0c4a4b4c23bcee2/F2.jpg", "media/e695c0c4a4b4c23bcee2/F3.jpg", "media/e695c0c4a4b4c23bcee2/F4.jpg", "media/e695c0c4a4b4c23bcee2/F5.jpg", "media/e695c0c4a4b4c23bcee2/F6.jpg", "media/e695c0c4a4b4c23bcee2/F7.jpg", "media/e695c0c4a4b4c23bcee2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5adaf22d00b4f0c47df3dde058404d40c9e088c354ea225bce3a736f1371266b"}
{"qa_id": "MOSE__4023fc77__object_world_motion_direction__01", "video_id": "MOSE__4023fc77", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (tall giraffe standing near the fence) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/acf5be44827fdf45e722/F0.jpg", "media/acf5be44827fdf45e722/F1.jpg", "media/acf5be44827fdf45e722/F2.jpg", "media/acf5be44827fdf45e722/F3.jpg", "media/acf5be44827fdf45e722/F4.jpg", "media/acf5be44827fdf45e722/F5.jpg", "media/acf5be44827fdf45e722/F6.jpg", "media/acf5be44827fdf45e722/F7.jpg", "media/acf5be44827fdf45e722/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (tall giraffe standing near the fence) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ae240add5eab6a697165a08d32d6f71c427eeb5980e908f8d6a2c5219d63cdcb"}
{"qa_id": "MOSE__405e3ee7__camera_view_direction_change__00", "video_id": "MOSE__405e3ee7", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/467a0431e20e44374504/F0.jpg", "media/467a0431e20e44374504/F1.jpg", "media/467a0431e20e44374504/F2.jpg", "media/467a0431e20e44374504/F3.jpg", "media/467a0431e20e44374504/F4.jpg", "media/467a0431e20e44374504/F5.jpg", "media/467a0431e20e44374504/F6.jpg", "media/467a0431e20e44374504/F7.jpg", "media/467a0431e20e44374504/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0395f32f9b844f8600c1e21bef7bc7a474df33e6384f9e952c050a4adc2fa1d0"}
{"qa_id": "MOSE__4411fdee__camera_object_distance_change__00", "video_id": "MOSE__4411fdee", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (green box truck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b4e38adc47eee0b93596/F0.jpg", "media/b4e38adc47eee0b93596/F1.jpg", "media/b4e38adc47eee0b93596/F2.jpg", "media/b4e38adc47eee0b93596/F3.jpg", "media/b4e38adc47eee0b93596/F4.jpg", "media/b4e38adc47eee0b93596/F5.jpg", "media/b4e38adc47eee0b93596/F6.jpg", "media/b4e38adc47eee0b93596/F7.jpg", "media/b4e38adc47eee0b93596/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (green box truck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c1a34f0973db88c5ddcf44314f31f1583104184b65f7b422b57914a3e9cb0f98"}
{"qa_id": "MOSE__4411fdee__camera_object_distance_change__01", "video_id": "MOSE__4411fdee", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (grey sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c5f5e34720e7d74fa924/F0.jpg", "media/c5f5e34720e7d74fa924/F1.jpg", "media/c5f5e34720e7d74fa924/F2.jpg", "media/c5f5e34720e7d74fa924/F3.jpg", "media/c5f5e34720e7d74fa924/F4.jpg", "media/c5f5e34720e7d74fa924/F5.jpg", "media/c5f5e34720e7d74fa924/F6.jpg", "media/c5f5e34720e7d74fa924/F7.jpg", "media/c5f5e34720e7d74fa924/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (grey sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e04527240b95b0b20484b72878b08bffd09e0b598a8ffe9b1fcb2b0b01bd9a99"}
{"qa_id": "MOSE__464066da__camera_translation_direction__00", "video_id": "MOSE__464066da", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "up", "forward", "down"], "frame_paths": ["media/09e6b6e6cfe7f33b5e0b/F0.jpg", "media/09e6b6e6cfe7f33b5e0b/F1.jpg", "media/09e6b6e6cfe7f33b5e0b/F2.jpg", "media/09e6b6e6cfe7f33b5e0b/F3.jpg", "media/09e6b6e6cfe7f33b5e0b/F4.jpg", "media/09e6b6e6cfe7f33b5e0b/F5.jpg", "media/09e6b6e6cfe7f33b5e0b/F6.jpg", "media/09e6b6e6cfe7f33b5e0b/F7.jpg", "media/09e6b6e6cfe7f33b5e0b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a813b28f1e4cf50680c994ab7ae4a74e5370fd07adf3bff41f15f6808e5977d"}
{"qa_id": "MOSE__46e00caf__object_world_motion_direction__00", "video_id": "MOSE__46e00caf", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange tabby cat with white paws) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/7a4a1fc4359de246ece0/F0.jpg", "media/7a4a1fc4359de246ece0/F1.jpg", "media/7a4a1fc4359de246ece0/F2.jpg", "media/7a4a1fc4359de246ece0/F3.jpg", "media/7a4a1fc4359de246ece0/F4.jpg", "media/7a4a1fc4359de246ece0/F5.jpg", "media/7a4a1fc4359de246ece0/F6.jpg", "media/7a4a1fc4359de246ece0/F7.jpg", "media/7a4a1fc4359de246ece0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange tabby cat with white paws) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3016c86670411a942d8088c8f52d089f802afa4743e04a260fee35b4e9e5cfb"}
{"qa_id": "MOSE__47891b4c__object_world_motion_direction__01", "video_id": "MOSE__47891b4c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (smaller juvenile elephant walking behind the adult) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "left", "up"], "frame_paths": ["media/b11c9a37345227107796/F0.jpg", "media/b11c9a37345227107796/F1.jpg", "media/b11c9a37345227107796/F2.jpg", "media/b11c9a37345227107796/F3.jpg", "media/b11c9a37345227107796/F4.jpg", "media/b11c9a37345227107796/F5.jpg", "media/b11c9a37345227107796/F6.jpg", "media/b11c9a37345227107796/F7.jpg", "media/b11c9a37345227107796/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (smaller juvenile elephant walking behind the adult) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c779ae4d654150dfb46bf1dc646576fc023a80f80196beacde04a358991d05c"}
{"qa_id": "MOSE__495be321__camera_view_direction_change__00", "video_id": "MOSE__495be321", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/073e44f5ae1088f5f958/F0.jpg", "media/073e44f5ae1088f5f958/F1.jpg", "media/073e44f5ae1088f5f958/F2.jpg", "media/073e44f5ae1088f5f958/F3.jpg", "media/073e44f5ae1088f5f958/F4.jpg", "media/073e44f5ae1088f5f958/F5.jpg", "media/073e44f5ae1088f5f958/F6.jpg", "media/073e44f5ae1088f5f958/F7.jpg", "media/073e44f5ae1088f5f958/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53eda269f6d9eb7cf97dc206d5e7093b2873b91fc6e510b9e39ce9e8cde770e"}
{"qa_id": "MOSE__4a41cbf8__object_world_motion_direction__00", "video_id": "MOSE__4a41cbf8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (large white fish with red mottling) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "right", "forward", "backward"], "frame_paths": ["media/1a53a689f6f67c2d1475/F0.jpg", "media/1a53a689f6f67c2d1475/F1.jpg", "media/1a53a689f6f67c2d1475/F2.jpg", "media/1a53a689f6f67c2d1475/F3.jpg", "media/1a53a689f6f67c2d1475/F4.jpg", "media/1a53a689f6f67c2d1475/F5.jpg", "media/1a53a689f6f67c2d1475/F6.jpg", "media/1a53a689f6f67c2d1475/F7.jpg", "media/1a53a689f6f67c2d1475/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (large white fish with red mottling) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b462827b89f1a1fcc185dddab453ca91acf5e4415f31d29b58cdce0f983d5281"}
{"qa_id": "MOSE__4a41cbf8__object_pair_distance_change__01", "video_id": "MOSE__4a41cbf8", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (small orange fish with dark spots) and object B (large white fish with red mottling) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8a574280b7ae5817573d/F0.jpg", "media/8a574280b7ae5817573d/F1.jpg", "media/8a574280b7ae5817573d/F2.jpg", "media/8a574280b7ae5817573d/F3.jpg", "media/8a574280b7ae5817573d/F4.jpg", "media/8a574280b7ae5817573d/F5.jpg", "media/8a574280b7ae5817573d/F6.jpg", "media/8a574280b7ae5817573d/F7.jpg", "media/8a574280b7ae5817573d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (small orange fish with dark spots) and object B (large white fish with red mottling) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d7569575a9ea7a7e0466d9851dd2085d1da1ca7c855cc9c55ef8f5ff7cb65bb8"}
{"qa_id": "MOSE__4c42c747__object_world_motion_direction__00", "video_id": "MOSE__4c42c747", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (small dark brown calf) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/1a077d16d62d39175d7c/F0.jpg", "media/1a077d16d62d39175d7c/F1.jpg", "media/1a077d16d62d39175d7c/F2.jpg", "media/1a077d16d62d39175d7c/F3.jpg", "media/1a077d16d62d39175d7c/F4.jpg", "media/1a077d16d62d39175d7c/F5.jpg", "media/1a077d16d62d39175d7c/F6.jpg", "media/1a077d16d62d39175d7c/F7.jpg", "media/1a077d16d62d39175d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (small dark brown calf) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e8a6b7be17f8575a1c298020aac056d2d363caf6fa9fd165b48fa2dccf778c6"}
{"qa_id": "MOSE__4c42c747__object_pair_distance_change__01", "video_id": "MOSE__4c42c747", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did the 3D distance between object A (large light brown cow with a halter) and object B (small dark brown calf) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f95983b261f9b367e823/F0.jpg", "media/f95983b261f9b367e823/F1.jpg", "media/f95983b261f9b367e823/F2.jpg", "media/f95983b261f9b367e823/F3.jpg", "media/f95983b261f9b367e823/F4.jpg", "media/f95983b261f9b367e823/F5.jpg", "media/f95983b261f9b367e823/F6.jpg", "media/f95983b261f9b367e823/F7.jpg", "media/f95983b261f9b367e823/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (large light brown cow with a halter) and object B (small dark brown calf) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d7ba54532abb497f3b68485b70188cf85264c7fac92d12f8c69d60f056e9682a"}
{"qa_id": "MOSE__4c42c747__object_travel_distance_comparison__02", "video_id": "MOSE__4c42c747", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large light brown cow with a halter) or B (small dark brown calf)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f95983b261f9b367e823/F0.jpg", "media/f95983b261f9b367e823/F1.jpg", "media/f95983b261f9b367e823/F2.jpg", "media/f95983b261f9b367e823/F3.jpg", "media/f95983b261f9b367e823/F4.jpg", "media/f95983b261f9b367e823/F5.jpg", "media/f95983b261f9b367e823/F6.jpg", "media/f95983b261f9b367e823/F7.jpg", "media/f95983b261f9b367e823/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large light brown cow with a halter) or B (small dark brown calf)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a8501368bb18487805bb45cf4702a4fedd0db51758d8baddd81df5764809e01"}
{"qa_id": "MOSE__4c94e4f3__camera_object_distance_change__00", "video_id": "MOSE__4c94e4f3", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (green box truck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/62d078ad5d821946d26e/F0.jpg", "media/62d078ad5d821946d26e/F1.jpg", "media/62d078ad5d821946d26e/F2.jpg", "media/62d078ad5d821946d26e/F3.jpg", "media/62d078ad5d821946d26e/F4.jpg", "media/62d078ad5d821946d26e/F5.jpg", "media/62d078ad5d821946d26e/F6.jpg", "media/62d078ad5d821946d26e/F7.jpg", "media/62d078ad5d821946d26e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (green box truck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2342546ebf0c1ba235f0430b49b941e50d62c7c2534fb4bc78a30c4abf0c7afa"}
{"qa_id": "MOSE__4e4483e7__object_world_motion_direction__00", "video_id": "MOSE__4e4483e7", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (sea turtle with patterned shell and flippers, visible on the right side) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "backward", "forward"], "frame_paths": ["media/05caea42158d84ed0479/F0.jpg", "media/05caea42158d84ed0479/F1.jpg", "media/05caea42158d84ed0479/F2.jpg", "media/05caea42158d84ed0479/F3.jpg", "media/05caea42158d84ed0479/F4.jpg", "media/05caea42158d84ed0479/F5.jpg", "media/05caea42158d84ed0479/F6.jpg", "media/05caea42158d84ed0479/F7.jpg", "media/05caea42158d84ed0479/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (sea turtle with patterned shell and flippers, visible on the right side) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c973807e42263fe14ecc0361ab9b4fd76ef30d323900f91a4be95ea4c305845"}
{"qa_id": "MOSE__4ed71139__camera_translation_direction__00", "video_id": "MOSE__4ed71139", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/e90d32c24eb33ce44db6/F0.jpg", "media/e90d32c24eb33ce44db6/F1.jpg", "media/e90d32c24eb33ce44db6/F2.jpg", "media/e90d32c24eb33ce44db6/F3.jpg", "media/e90d32c24eb33ce44db6/F4.jpg", "media/e90d32c24eb33ce44db6/F5.jpg", "media/e90d32c24eb33ce44db6/F6.jpg", "media/e90d32c24eb33ce44db6/F7.jpg", "media/e90d32c24eb33ce44db6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f74d5eadb4f59dbfd62079e4b8d9cd16efadec5e8453c2a41ee58fe2192962d"}
{"qa_id": "MOSE__4ed71139__camera_view_direction_change__01", "video_id": "MOSE__4ed71139", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/e90d32c24eb33ce44db6/F0.jpg", "media/e90d32c24eb33ce44db6/F1.jpg", "media/e90d32c24eb33ce44db6/F2.jpg", "media/e90d32c24eb33ce44db6/F3.jpg", "media/e90d32c24eb33ce44db6/F4.jpg", "media/e90d32c24eb33ce44db6/F5.jpg", "media/e90d32c24eb33ce44db6/F6.jpg", "media/e90d32c24eb33ce44db6/F7.jpg", "media/e90d32c24eb33ce44db6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "MOSE__4ed71139__camera_object_distance_change__02", "video_id": "MOSE__4ed71139", "task": "camera_object_distance_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, did object A (large ferry with 'HELLENIC SEAWAYS' written on the side) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d225d46a811a754ca657/F0.jpg", "media/d225d46a811a754ca657/F1.jpg", "media/d225d46a811a754ca657/F2.jpg", "media/d225d46a811a754ca657/F3.jpg", "media/d225d46a811a754ca657/F4.jpg", "media/d225d46a811a754ca657/F5.jpg", "media/d225d46a811a754ca657/F6.jpg", "media/d225d46a811a754ca657/F7.jpg", "media/d225d46a811a754ca657/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (large ferry with 'HELLENIC SEAWAYS' written on the side) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82220ed98fe35f7e9874da8fad91af3feeb5f62c62f4576b01f0fed7461bbd85"}
{"qa_id": "MOSE__4f5db794__camera_translation_direction__00", "video_id": "MOSE__4f5db794", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "right", "left", "down"], "frame_paths": ["media/a59738fce215c5ce1453/F0.jpg", "media/a59738fce215c5ce1453/F1.jpg", "media/a59738fce215c5ce1453/F2.jpg", "media/a59738fce215c5ce1453/F3.jpg", "media/a59738fce215c5ce1453/F4.jpg", "media/a59738fce215c5ce1453/F5.jpg", "media/a59738fce215c5ce1453/F6.jpg", "media/a59738fce215c5ce1453/F7.jpg", "media/a59738fce215c5ce1453/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "161629e5c6562d9a17075fd2761c76b8c943a242f9d31535a5354f1d3bfde808"}
{"qa_id": "MOSE__4f5db794__object_world_motion_direction__01", "video_id": "MOSE__4f5db794", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (dark bird with mottled brown and black feathers perched on a wooden dowel) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "down", "left", "forward"], "frame_paths": ["media/45884f06088e43dbe119/F0.jpg", "media/45884f06088e43dbe119/F1.jpg", "media/45884f06088e43dbe119/F2.jpg", "media/45884f06088e43dbe119/F3.jpg", "media/45884f06088e43dbe119/F4.jpg", "media/45884f06088e43dbe119/F5.jpg", "media/45884f06088e43dbe119/F6.jpg", "media/45884f06088e43dbe119/F7.jpg", "media/45884f06088e43dbe119/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (dark bird with mottled brown and black feathers perched on a wooden dowel) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d48aad59cc861fe9048adcc2050e0fc7121dc001aee67c735447898883461a2"}
{"qa_id": "MOSE__51586b36__camera_object_distance_change__00", "video_id": "MOSE__51586b36", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f042ad6f36df5f032551/F0.jpg", "media/f042ad6f36df5f032551/F1.jpg", "media/f042ad6f36df5f032551/F2.jpg", "media/f042ad6f36df5f032551/F3.jpg", "media/f042ad6f36df5f032551/F4.jpg", "media/f042ad6f36df5f032551/F5.jpg", "media/f042ad6f36df5f032551/F6.jpg", "media/f042ad6f36df5f032551/F7.jpg", "media/f042ad6f36df5f032551/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66fa232ba25e1148feb78d310b291269a7a706f0d5215db17fb6047f0b5a27e3"}
{"qa_id": "MOSE__51586b36__camera_object_distance_change__01", "video_id": "MOSE__51586b36", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object B (person in black shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/80449ec51ba55e0eff5d/F0.jpg", "media/80449ec51ba55e0eff5d/F1.jpg", "media/80449ec51ba55e0eff5d/F2.jpg", "media/80449ec51ba55e0eff5d/F3.jpg", "media/80449ec51ba55e0eff5d/F4.jpg", "media/80449ec51ba55e0eff5d/F5.jpg", "media/80449ec51ba55e0eff5d/F6.jpg", "media/80449ec51ba55e0eff5d/F7.jpg", "media/80449ec51ba55e0eff5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (person in black shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1d04171e4b9d3d999c274d51a3b1c22b737b9a9124d36b723b15e12d575f5c6"}
{"qa_id": "MOSE__51befc31__object_world_motion_direction__00", "video_id": "MOSE__51befc31", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object A (large brown turtle with a smooth, light-colored shell) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/66e97274522d267b912b/F0.jpg", "media/66e97274522d267b912b/F1.jpg", "media/66e97274522d267b912b/F2.jpg", "media/66e97274522d267b912b/F3.jpg", "media/66e97274522d267b912b/F4.jpg", "media/66e97274522d267b912b/F5.jpg", "media/66e97274522d267b912b/F6.jpg", "media/66e97274522d267b912b/F7.jpg", "media/66e97274522d267b912b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object A (large brown turtle with a smooth, light-colored shell) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "41c9377116693e2ea0978e95facdecf2ca6d980924c8a5d86cb37449170e7c80"}
{"qa_id": "MOSE__51befc31__object_travel_distance_comparison__01", "video_id": "MOSE__51befc31", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large brown turtle with a smooth, light-colored shell) or B (dark-shelled turtle swimming near the surface)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/10cd9aed0fa67d91572a/F0.jpg", "media/10cd9aed0fa67d91572a/F1.jpg", "media/10cd9aed0fa67d91572a/F2.jpg", "media/10cd9aed0fa67d91572a/F3.jpg", "media/10cd9aed0fa67d91572a/F4.jpg", "media/10cd9aed0fa67d91572a/F5.jpg", "media/10cd9aed0fa67d91572a/F6.jpg", "media/10cd9aed0fa67d91572a/F7.jpg", "media/10cd9aed0fa67d91572a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large brown turtle with a smooth, light-colored shell) or B (dark-shelled turtle swimming near the surface)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c5195c2f6e6ed48b239e4da54e1a0b3cd1707f15e59a9f27ff686eacc3c5af3"}
{"qa_id": "MOSE__529c6889__object_world_motion_direction__00", "video_id": "MOSE__529c6889", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object B (white sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "down", "right"], "frame_paths": ["media/7492418c6862f02670b1/F0.jpg", "media/7492418c6862f02670b1/F1.jpg", "media/7492418c6862f02670b1/F2.jpg", "media/7492418c6862f02670b1/F3.jpg", "media/7492418c6862f02670b1/F4.jpg", "media/7492418c6862f02670b1/F5.jpg", "media/7492418c6862f02670b1/F6.jpg", "media/7492418c6862f02670b1/F7.jpg", "media/7492418c6862f02670b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object B (white sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "583d83fc3474e6f3a707b853fcb10e2bf99f42c7b88b110c7066b6e184f141c5"}
{"qa_id": "MOSE__529c6889__object_pair_distance_change__01", "video_id": "MOSE__529c6889", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (handlebar of a vehicle with a black cloth tied to it) and object B (white sedan) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/bcf3f5235b813c729654/F0.jpg", "media/bcf3f5235b813c729654/F1.jpg", "media/bcf3f5235b813c729654/F2.jpg", "media/bcf3f5235b813c729654/F3.jpg", "media/bcf3f5235b813c729654/F4.jpg", "media/bcf3f5235b813c729654/F5.jpg", "media/bcf3f5235b813c729654/F6.jpg", "media/bcf3f5235b813c729654/F7.jpg", "media/bcf3f5235b813c729654/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (handlebar of a vehicle with a black cloth tied to it) and object B (white sedan) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3c670e5e8c13bfd22ff62a3609dd3326de64540606e46e5c50457cd7d4206ff1"}
{"qa_id": "MOSE__529c6889__object_travel_distance_comparison__02", "video_id": "MOSE__529c6889", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (handlebar of a vehicle with a black cloth tied to it) or B (white sedan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/bcf3f5235b813c729654/F0.jpg", "media/bcf3f5235b813c729654/F1.jpg", "media/bcf3f5235b813c729654/F2.jpg", "media/bcf3f5235b813c729654/F3.jpg", "media/bcf3f5235b813c729654/F4.jpg", "media/bcf3f5235b813c729654/F5.jpg", "media/bcf3f5235b813c729654/F6.jpg", "media/bcf3f5235b813c729654/F7.jpg", "media/bcf3f5235b813c729654/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (handlebar of a vehicle with a black cloth tied to it) or B (white sedan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "872be652e0a7577e954b7c9362c2ae245d9c442f072499fbfeaf6b5e8628be34"}
{"qa_id": "MOSE__52ddf44c__object_world_motion_direction__00", "video_id": "MOSE__52ddf44c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown chicken standing on the left) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/663ebcb3ad46fb2ff3c4/F0.jpg", "media/663ebcb3ad46fb2ff3c4/F1.jpg", "media/663ebcb3ad46fb2ff3c4/F2.jpg", "media/663ebcb3ad46fb2ff3c4/F3.jpg", "media/663ebcb3ad46fb2ff3c4/F4.jpg", "media/663ebcb3ad46fb2ff3c4/F5.jpg", "media/663ebcb3ad46fb2ff3c4/F6.jpg", "media/663ebcb3ad46fb2ff3c4/F7.jpg", "media/663ebcb3ad46fb2ff3c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown chicken standing on the left) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d50cda02ce8340c06261d0b4b0871b18f17ea4fc4d502c7cdffb6acc78f9db14"}
{"qa_id": "MOSE__52ddf44c__object_pair_distance_change__01", "video_id": "MOSE__52ddf44c", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (light brown chicken standing on the left) and object B (light brown chicken standing in the center) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/6c593fa0c2a2036e4ef7/F0.jpg", "media/6c593fa0c2a2036e4ef7/F1.jpg", "media/6c593fa0c2a2036e4ef7/F2.jpg", "media/6c593fa0c2a2036e4ef7/F3.jpg", "media/6c593fa0c2a2036e4ef7/F4.jpg", "media/6c593fa0c2a2036e4ef7/F5.jpg", "media/6c593fa0c2a2036e4ef7/F6.jpg", "media/6c593fa0c2a2036e4ef7/F7.jpg", "media/6c593fa0c2a2036e4ef7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (light brown chicken standing on the left) and object B (light brown chicken standing in the center) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c8731629d0a8ec6b5f3124dab9c1eee00eddb638b6546c17ea3d00c1b310b116"}
{"qa_id": "MOSE__52ddf44c__object_travel_distance_comparison__02", "video_id": "MOSE__52ddf44c", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown chicken standing on the left) or B (light brown chicken standing in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/6c593fa0c2a2036e4ef7/F0.jpg", "media/6c593fa0c2a2036e4ef7/F1.jpg", "media/6c593fa0c2a2036e4ef7/F2.jpg", "media/6c593fa0c2a2036e4ef7/F3.jpg", "media/6c593fa0c2a2036e4ef7/F4.jpg", "media/6c593fa0c2a2036e4ef7/F5.jpg", "media/6c593fa0c2a2036e4ef7/F6.jpg", "media/6c593fa0c2a2036e4ef7/F7.jpg", "media/6c593fa0c2a2036e4ef7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown chicken standing on the left) or B (light brown chicken standing in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbe9474b01e107a5505955b5675d5528f0368fdb8d5fe5acf7701dbd7b193570"}
{"qa_id": "MOSE__52fa9174__camera_translation_direction__00", "video_id": "MOSE__52fa9174", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "backward", "up", "down"], "frame_paths": ["media/78870ff2d3b26b176aff/F0.jpg", "media/78870ff2d3b26b176aff/F1.jpg", "media/78870ff2d3b26b176aff/F2.jpg", "media/78870ff2d3b26b176aff/F3.jpg", "media/78870ff2d3b26b176aff/F4.jpg", "media/78870ff2d3b26b176aff/F5.jpg", "media/78870ff2d3b26b176aff/F6.jpg", "media/78870ff2d3b26b176aff/F7.jpg", "media/78870ff2d3b26b176aff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c8378357a25db80017060cba3274129d96d5a2f67c5976f5921bcdef9e7adbc5"}
{"qa_id": "MOSE__52fa9174__object_world_motion_direction__01", "video_id": "MOSE__52fa9174", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (orange and white kitten in the center) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "down", "up"], "frame_paths": ["media/60b66da997e088aad163/F0.jpg", "media/60b66da997e088aad163/F1.jpg", "media/60b66da997e088aad163/F2.jpg", "media/60b66da997e088aad163/F3.jpg", "media/60b66da997e088aad163/F4.jpg", "media/60b66da997e088aad163/F5.jpg", "media/60b66da997e088aad163/F6.jpg", "media/60b66da997e088aad163/F7.jpg", "media/60b66da997e088aad163/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (orange and white kitten in the center) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8af1e4cca2af7efdc9a36f2c43594c59b4025f6ce2bdec90299c4168bbc6d6b8"}
{"qa_id": "MOSE__52fa9174__camera_object_distance_change__02", "video_id": "MOSE__52fa9174", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (orange and white kitten in the center) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/60b66da997e088aad163/F0.jpg", "media/60b66da997e088aad163/F1.jpg", "media/60b66da997e088aad163/F2.jpg", "media/60b66da997e088aad163/F3.jpg", "media/60b66da997e088aad163/F4.jpg", "media/60b66da997e088aad163/F5.jpg", "media/60b66da997e088aad163/F6.jpg", "media/60b66da997e088aad163/F7.jpg", "media/60b66da997e088aad163/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (orange and white kitten in the center) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c6e50714d12098d35a902ad03d80d81410ecf396bb911f57de76bf6071cbdc97"}
{"qa_id": "MOSE__52fa9174__object_pair_distance_change__03", "video_id": "MOSE__52fa9174", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did the 3D distance between object A (orange and white kitten on the left) and object B (orange and white kitten in the center) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/0bf17736ec75395b22ac/F0.jpg", "media/0bf17736ec75395b22ac/F1.jpg", "media/0bf17736ec75395b22ac/F2.jpg", "media/0bf17736ec75395b22ac/F3.jpg", "media/0bf17736ec75395b22ac/F4.jpg", "media/0bf17736ec75395b22ac/F5.jpg", "media/0bf17736ec75395b22ac/F6.jpg", "media/0bf17736ec75395b22ac/F7.jpg", "media/0bf17736ec75395b22ac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (orange and white kitten on the left) and object B (orange and white kitten in the center) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b9cca85cfac4936c3d8633acb35f1eec213ddb2f95aee47699164eb69995d04"}
{"qa_id": "MOSE__5411113f__object_world_motion_direction__00", "video_id": "MOSE__5411113f", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (orange electric scooter with a purple patterned windbreaker quilt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "forward", "up", "down", "right"], "frame_paths": ["media/72fc496c2a0643c3aeac/F0.jpg", "media/72fc496c2a0643c3aeac/F1.jpg", "media/72fc496c2a0643c3aeac/F2.jpg", "media/72fc496c2a0643c3aeac/F3.jpg", "media/72fc496c2a0643c3aeac/F4.jpg", "media/72fc496c2a0643c3aeac/F5.jpg", "media/72fc496c2a0643c3aeac/F6.jpg", "media/72fc496c2a0643c3aeac/F7.jpg", "media/72fc496c2a0643c3aeac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (orange electric scooter with a purple patterned windbreaker quilt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. forward\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "43fc1869f576cb3eab2f4b10bcfbf5e3b344784bb838306f55874920bdffbfd9"}
{"qa_id": "MOSE__5411113f__camera_object_distance_change__01", "video_id": "MOSE__5411113f", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object A (orange electric scooter with a purple patterned windbreaker quilt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/72fc496c2a0643c3aeac/F0.jpg", "media/72fc496c2a0643c3aeac/F1.jpg", "media/72fc496c2a0643c3aeac/F2.jpg", "media/72fc496c2a0643c3aeac/F3.jpg", "media/72fc496c2a0643c3aeac/F4.jpg", "media/72fc496c2a0643c3aeac/F5.jpg", "media/72fc496c2a0643c3aeac/F6.jpg", "media/72fc496c2a0643c3aeac/F7.jpg", "media/72fc496c2a0643c3aeac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (orange electric scooter with a purple patterned windbreaker quilt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca2d719029479968eeca3d0a2c3e575a08a01954fa6477c198a2c2041d6be107"}
{"qa_id": "MOSE__5411113f__object_world_motion_direction__02", "video_id": "MOSE__5411113f", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object B (person wearing a white helmet and black clothing riding the scooter) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/577f472ee80b673ec353/F0.jpg", "media/577f472ee80b673ec353/F1.jpg", "media/577f472ee80b673ec353/F2.jpg", "media/577f472ee80b673ec353/F3.jpg", "media/577f472ee80b673ec353/F4.jpg", "media/577f472ee80b673ec353/F5.jpg", "media/577f472ee80b673ec353/F6.jpg", "media/577f472ee80b673ec353/F7.jpg", "media/577f472ee80b673ec353/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object B (person wearing a white helmet and black clothing riding the scooter) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eb2e6f4fed3ccf660bc82f562fbe06a98cec01bf3a1c9f6a0d3295ca0b1293a3"}
{"qa_id": "MOSE__5411113f__camera_object_distance_change__03", "video_id": "MOSE__5411113f", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object B (person wearing a white helmet and black clothing riding the scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/577f472ee80b673ec353/F0.jpg", "media/577f472ee80b673ec353/F1.jpg", "media/577f472ee80b673ec353/F2.jpg", "media/577f472ee80b673ec353/F3.jpg", "media/577f472ee80b673ec353/F4.jpg", "media/577f472ee80b673ec353/F5.jpg", "media/577f472ee80b673ec353/F6.jpg", "media/577f472ee80b673ec353/F7.jpg", "media/577f472ee80b673ec353/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object B (person wearing a white helmet and black clothing riding the scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c96855309bd3ca7cb43fe2a4f09b53d8ac14ea3765461c218164f07ef4b2f2c"}
{"qa_id": "MOSE__5416fbb7__camera_translation_direction__00", "video_id": "MOSE__5416fbb7", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/c2ecd6bb96d3338740ef/F0.jpg", "media/c2ecd6bb96d3338740ef/F1.jpg", "media/c2ecd6bb96d3338740ef/F2.jpg", "media/c2ecd6bb96d3338740ef/F3.jpg", "media/c2ecd6bb96d3338740ef/F4.jpg", "media/c2ecd6bb96d3338740ef/F5.jpg", "media/c2ecd6bb96d3338740ef/F6.jpg", "media/c2ecd6bb96d3338740ef/F7.jpg", "media/c2ecd6bb96d3338740ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84d72235cb5e1ffe9c608fb14bdbfe4c7e64cdc9fe86a9c032851002de63d6f1"}
{"qa_id": "MOSE__54273925__camera_view_direction_change__00", "video_id": "MOSE__54273925", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/7b170d874f4dfaa64f48/F0.jpg", "media/7b170d874f4dfaa64f48/F1.jpg", "media/7b170d874f4dfaa64f48/F2.jpg", "media/7b170d874f4dfaa64f48/F3.jpg", "media/7b170d874f4dfaa64f48/F4.jpg", "media/7b170d874f4dfaa64f48/F5.jpg", "media/7b170d874f4dfaa64f48/F6.jpg", "media/7b170d874f4dfaa64f48/F7.jpg", "media/7b170d874f4dfaa64f48/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e50e3389ddfccaab5273ba60086b3d0ca2a36a5b2b0841845ad6dcc615cd56e8"}
{"qa_id": "MOSE__54364322__camera_object_distance_change__00", "video_id": "MOSE__54364322", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b7502ba8d24394f277ff/F0.jpg", "media/b7502ba8d24394f277ff/F1.jpg", "media/b7502ba8d24394f277ff/F2.jpg", "media/b7502ba8d24394f277ff/F3.jpg", "media/b7502ba8d24394f277ff/F4.jpg", "media/b7502ba8d24394f277ff/F5.jpg", "media/b7502ba8d24394f277ff/F6.jpg", "media/b7502ba8d24394f277ff/F7.jpg", "media/b7502ba8d24394f277ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5eac1b1d61c9b699babc5d3c3211ea5b3a32fb87e0d5727ff6eff507145aa8a"}
{"qa_id": "MOSE__54364322__object_pair_distance_change__01", "video_id": "MOSE__54364322", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (white SUV) and object B (white SUV) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/5cfaf0125bca3160e4dc/F0.jpg", "media/5cfaf0125bca3160e4dc/F1.jpg", "media/5cfaf0125bca3160e4dc/F2.jpg", "media/5cfaf0125bca3160e4dc/F3.jpg", "media/5cfaf0125bca3160e4dc/F4.jpg", "media/5cfaf0125bca3160e4dc/F5.jpg", "media/5cfaf0125bca3160e4dc/F6.jpg", "media/5cfaf0125bca3160e4dc/F7.jpg", "media/5cfaf0125bca3160e4dc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (white SUV) and object B (white SUV) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "76b0cea9fefc85ec87fe2789e7ac0201601b28a189acee813f864d613de04b3c"}
{"qa_id": "MOSE__5457b298__camera_view_direction_change__00", "video_id": "MOSE__5457b298", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/88ad5c8448513c0b3f28/F0.jpg", "media/88ad5c8448513c0b3f28/F1.jpg", "media/88ad5c8448513c0b3f28/F2.jpg", "media/88ad5c8448513c0b3f28/F3.jpg", "media/88ad5c8448513c0b3f28/F4.jpg", "media/88ad5c8448513c0b3f28/F5.jpg", "media/88ad5c8448513c0b3f28/F6.jpg", "media/88ad5c8448513c0b3f28/F7.jpg", "media/88ad5c8448513c0b3f28/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da4720e6165f7bcd1ab67d5bd15b161a9966096a666974b31ff1b4cece130e0e"}
{"qa_id": "MOSE__5457b298__camera_object_distance_change__02", "video_id": "MOSE__5457b298", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, did object A (black motorcycle with a rider in a dark blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/4bf391e296875f9c0111/F0.jpg", "media/4bf391e296875f9c0111/F1.jpg", "media/4bf391e296875f9c0111/F2.jpg", "media/4bf391e296875f9c0111/F3.jpg", "media/4bf391e296875f9c0111/F4.jpg", "media/4bf391e296875f9c0111/F5.jpg", "media/4bf391e296875f9c0111/F6.jpg", "media/4bf391e296875f9c0111/F7.jpg", "media/4bf391e296875f9c0111/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, did object A (black motorcycle with a rider in a dark blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36f8f6eeff07beb2564fc64f15376e60594470b41f3d81c88a366f3a7cc1e19b"}
{"qa_id": "MOSE__5457b298__object_world_motion_direction__03", "video_id": "MOSE__5457b298", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F7, in which dominant direction did object B (black motorcycle with a rider in a dark grey shirt) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "left", "down", "right"], "frame_paths": ["media/34f2d2b0fd67475319de/F0.jpg", "media/34f2d2b0fd67475319de/F1.jpg", "media/34f2d2b0fd67475319de/F2.jpg", "media/34f2d2b0fd67475319de/F3.jpg", "media/34f2d2b0fd67475319de/F4.jpg", "media/34f2d2b0fd67475319de/F5.jpg", "media/34f2d2b0fd67475319de/F6.jpg", "media/34f2d2b0fd67475319de/F7.jpg", "media/34f2d2b0fd67475319de/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F7, in which dominant direction did object B (black motorcycle with a rider in a dark grey shirt) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cd7491f0ef12d9280de1fb8bea20f5924c804391403fe2a3bf44e99f85cc7cb2"}
{"qa_id": "MOSE__54b8f502__camera_translation_direction__00", "video_id": "MOSE__54b8f502", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/9167cf694763eacfbf06/F0.jpg", "media/9167cf694763eacfbf06/F1.jpg", "media/9167cf694763eacfbf06/F2.jpg", "media/9167cf694763eacfbf06/F3.jpg", "media/9167cf694763eacfbf06/F4.jpg", "media/9167cf694763eacfbf06/F5.jpg", "media/9167cf694763eacfbf06/F6.jpg", "media/9167cf694763eacfbf06/F7.jpg", "media/9167cf694763eacfbf06/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44531bdadabea5c7198704413dd43722379378ca88ef6b4b092016944e2368eb"}
{"qa_id": "MOSE__54b9dbba__object_world_motion_direction__00", "video_id": "MOSE__54b9dbba", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (light brown bear cub) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "backward", "left", "right"], "frame_paths": ["media/056de7cbedd0f497f8d4/F0.jpg", "media/056de7cbedd0f497f8d4/F1.jpg", "media/056de7cbedd0f497f8d4/F2.jpg", "media/056de7cbedd0f497f8d4/F3.jpg", "media/056de7cbedd0f497f8d4/F4.jpg", "media/056de7cbedd0f497f8d4/F5.jpg", "media/056de7cbedd0f497f8d4/F6.jpg", "media/056de7cbedd0f497f8d4/F7.jpg", "media/056de7cbedd0f497f8d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (light brown bear cub) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "453a8f58fba61a912d0d1089aaf5f628f8a2b58194548f086e8412f035f85c2e"}
{"qa_id": "MOSE__54b9dbba__camera_object_distance_change__01", "video_id": "MOSE__54b9dbba", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (light brown bear cub) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/056de7cbedd0f497f8d4/F0.jpg", "media/056de7cbedd0f497f8d4/F1.jpg", "media/056de7cbedd0f497f8d4/F2.jpg", "media/056de7cbedd0f497f8d4/F3.jpg", "media/056de7cbedd0f497f8d4/F4.jpg", "media/056de7cbedd0f497f8d4/F5.jpg", "media/056de7cbedd0f497f8d4/F6.jpg", "media/056de7cbedd0f497f8d4/F7.jpg", "media/056de7cbedd0f497f8d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (light brown bear cub) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3b536753f63bcc7d6aa34db391f07d8fdc5b53c9d274f72b7bdd1023b711a5ae"}
{"qa_id": "MOSE__54b9dbba__object_pair_distance_change__02", "video_id": "MOSE__54b9dbba", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (light brown bear cub) and object B (dark brown bear cub) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/13ab5690b289953b884e/F0.jpg", "media/13ab5690b289953b884e/F1.jpg", "media/13ab5690b289953b884e/F2.jpg", "media/13ab5690b289953b884e/F3.jpg", "media/13ab5690b289953b884e/F4.jpg", "media/13ab5690b289953b884e/F5.jpg", "media/13ab5690b289953b884e/F6.jpg", "media/13ab5690b289953b884e/F7.jpg", "media/13ab5690b289953b884e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (light brown bear cub) and object B (dark brown bear cub) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "882987b71975cfca66e2d67ab9eab72439e67170d027a43200d004e0076dcf06"}
{"qa_id": "MOSE__54b9dbba__object_travel_distance_comparison__03", "video_id": "MOSE__54b9dbba", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown bear cub) or B (dark brown bear cub)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/13ab5690b289953b884e/F0.jpg", "media/13ab5690b289953b884e/F1.jpg", "media/13ab5690b289953b884e/F2.jpg", "media/13ab5690b289953b884e/F3.jpg", "media/13ab5690b289953b884e/F4.jpg", "media/13ab5690b289953b884e/F5.jpg", "media/13ab5690b289953b884e/F6.jpg", "media/13ab5690b289953b884e/F7.jpg", "media/13ab5690b289953b884e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown bear cub) or B (dark brown bear cub)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca5987bbb243376190412ff07ecdb1a958144105db0b6ab716e636a395d9d4e8"}
{"qa_id": "MOSE__55ae6921__object_world_motion_direction__00", "video_id": "MOSE__55ae6921", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in black outfit) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/74e1f0ff79b1e4c08c89/F0.jpg", "media/74e1f0ff79b1e4c08c89/F1.jpg", "media/74e1f0ff79b1e4c08c89/F2.jpg", "media/74e1f0ff79b1e4c08c89/F3.jpg", "media/74e1f0ff79b1e4c08c89/F4.jpg", "media/74e1f0ff79b1e4c08c89/F5.jpg", "media/74e1f0ff79b1e4c08c89/F6.jpg", "media/74e1f0ff79b1e4c08c89/F7.jpg", "media/74e1f0ff79b1e4c08c89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in black outfit) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d86a76a577732898160241942b2be54205fdad61ba5059723e7caa1c798b01cc"}
{"qa_id": "MOSE__56a5643a__object_world_motion_direction__00", "video_id": "MOSE__56a5643a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "up", "forward", "down"], "frame_paths": ["media/f356084b3634a3c766f8/F0.jpg", "media/f356084b3634a3c766f8/F1.jpg", "media/f356084b3634a3c766f8/F2.jpg", "media/f356084b3634a3c766f8/F3.jpg", "media/f356084b3634a3c766f8/F4.jpg", "media/f356084b3634a3c766f8/F5.jpg", "media/f356084b3634a3c766f8/F6.jpg", "media/f356084b3634a3c766f8/F7.jpg", "media/f356084b3634a3c766f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb9674efb31f6ea6dd73bc547786d1b622d3a55edaadac79e26c8b26842df477"}
{"qa_id": "MOSE__56b9ce41__object_world_motion_direction__00", "video_id": "MOSE__56b9ce41", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "left", "right", "backward"], "frame_paths": ["media/b68159f479af0b361b99/F0.jpg", "media/b68159f479af0b361b99/F1.jpg", "media/b68159f479af0b361b99/F2.jpg", "media/b68159f479af0b361b99/F3.jpg", "media/b68159f479af0b361b99/F4.jpg", "media/b68159f479af0b361b99/F5.jpg", "media/b68159f479af0b361b99/F6.jpg", "media/b68159f479af0b361b99/F7.jpg", "media/b68159f479af0b361b99/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8ba04548b536138e12c782abb166dfac9133877db835b878cba5de499ca8dcd"}
{"qa_id": "MOSE__57ad022e__object_world_motion_direction__00", "video_id": "MOSE__57ad022e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (tabby cat on the left) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "right", "up", "backward"], "frame_paths": ["media/bfdb7fff48269c8ca4a7/F0.jpg", "media/bfdb7fff48269c8ca4a7/F1.jpg", "media/bfdb7fff48269c8ca4a7/F2.jpg", "media/bfdb7fff48269c8ca4a7/F3.jpg", "media/bfdb7fff48269c8ca4a7/F4.jpg", "media/bfdb7fff48269c8ca4a7/F5.jpg", "media/bfdb7fff48269c8ca4a7/F6.jpg", "media/bfdb7fff48269c8ca4a7/F7.jpg", "media/bfdb7fff48269c8ca4a7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (tabby cat on the left) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d14e9beb505d451cd0ecb2b0f1bf1284e0e4b49f24cd0facb63794062b4295c9"}
{"qa_id": "MOSE__57f3a53e__object_world_motion_direction__00", "video_id": "MOSE__57f3a53e", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown chicken standing near wooden posts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "left", "right", "backward"], "frame_paths": ["media/2e586801810f8270aa85/F0.jpg", "media/2e586801810f8270aa85/F1.jpg", "media/2e586801810f8270aa85/F2.jpg", "media/2e586801810f8270aa85/F3.jpg", "media/2e586801810f8270aa85/F4.jpg", "media/2e586801810f8270aa85/F5.jpg", "media/2e586801810f8270aa85/F6.jpg", "media/2e586801810f8270aa85/F7.jpg", "media/2e586801810f8270aa85/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown chicken standing near wooden posts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "41651eb5f1c19fb6c2640b971b1d2612c0f16d996c98ec607674e133c401951d"}
{"qa_id": "MOSE__57f3a53e__object_travel_distance_comparison__01", "video_id": "MOSE__57f3a53e", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown chicken standing near wooden posts) or B (brown chicken pecking at the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/9f9eeea5a8ae4ee86151/F0.jpg", "media/9f9eeea5a8ae4ee86151/F1.jpg", "media/9f9eeea5a8ae4ee86151/F2.jpg", "media/9f9eeea5a8ae4ee86151/F3.jpg", "media/9f9eeea5a8ae4ee86151/F4.jpg", "media/9f9eeea5a8ae4ee86151/F5.jpg", "media/9f9eeea5a8ae4ee86151/F6.jpg", "media/9f9eeea5a8ae4ee86151/F7.jpg", "media/9f9eeea5a8ae4ee86151/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown chicken standing near wooden posts) or B (brown chicken pecking at the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d6de52959fee50bbe0323b6a21c78ec593e3da9b70fb05746e87921cd2ad2632"}
{"qa_id": "MOSE__5845c157__object_world_motion_direction__00", "video_id": "MOSE__5845c157", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (zebra running through the water on the right side of the group) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "backward", "forward", "left", "up"], "frame_paths": ["media/2bf23ce972759e5c6c9f/F0.jpg", "media/2bf23ce972759e5c6c9f/F1.jpg", "media/2bf23ce972759e5c6c9f/F2.jpg", "media/2bf23ce972759e5c6c9f/F3.jpg", "media/2bf23ce972759e5c6c9f/F4.jpg", "media/2bf23ce972759e5c6c9f/F5.jpg", "media/2bf23ce972759e5c6c9f/F6.jpg", "media/2bf23ce972759e5c6c9f/F7.jpg", "media/2bf23ce972759e5c6c9f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (zebra running through the water on the right side of the group) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. backward\nD. forward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "adc39ee264872ceb0aa196539f521c76adef6114d7b789fbb269af58cbf7e0ae"}
{"qa_id": "MOSE__584838d4__camera_object_distance_change__00", "video_id": "MOSE__584838d4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/36bbbe37e84c518ca288/F0.jpg", "media/36bbbe37e84c518ca288/F1.jpg", "media/36bbbe37e84c518ca288/F2.jpg", "media/36bbbe37e84c518ca288/F3.jpg", "media/36bbbe37e84c518ca288/F4.jpg", "media/36bbbe37e84c518ca288/F5.jpg", "media/36bbbe37e84c518ca288/F6.jpg", "media/36bbbe37e84c518ca288/F7.jpg", "media/36bbbe37e84c518ca288/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66fa232ba25e1148feb78d310b291269a7a706f0d5215db17fb6047f0b5a27e3"}
{"qa_id": "MOSE__584838d4__camera_object_distance_change__01", "video_id": "MOSE__584838d4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object B (person in dark shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c2078efd7173a7a7ff88/F0.jpg", "media/c2078efd7173a7a7ff88/F1.jpg", "media/c2078efd7173a7a7ff88/F2.jpg", "media/c2078efd7173a7a7ff88/F3.jpg", "media/c2078efd7173a7a7ff88/F4.jpg", "media/c2078efd7173a7a7ff88/F5.jpg", "media/c2078efd7173a7a7ff88/F6.jpg", "media/c2078efd7173a7a7ff88/F7.jpg", "media/c2078efd7173a7a7ff88/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (person in dark shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a65973e878a1eae6853c139199a067ed20c7a6c754a0098a42327da2ba58aa52"}
{"qa_id": "MOSE__58f24fa4__object_world_motion_direction__00", "video_id": "MOSE__58f24fa4", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (brown bull with a large hump and horns, facing right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/36c99262e22c3e390a5f/F0.jpg", "media/36c99262e22c3e390a5f/F1.jpg", "media/36c99262e22c3e390a5f/F2.jpg", "media/36c99262e22c3e390a5f/F3.jpg", "media/36c99262e22c3e390a5f/F4.jpg", "media/36c99262e22c3e390a5f/F5.jpg", "media/36c99262e22c3e390a5f/F6.jpg", "media/36c99262e22c3e390a5f/F7.jpg", "media/36c99262e22c3e390a5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (brown bull with a large hump and horns, facing right) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "932c30d4c9fe1850f9cc0d4bc98439668a0f9551fc32474b8275da8c6c1f2c6e"}
{"qa_id": "MOSE__58f24fa4__object_world_motion_direction__01", "video_id": "MOSE__58f24fa4", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (brown bull with white markings on its back, facing left) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/82b7cf0b81e19760af38/F0.jpg", "media/82b7cf0b81e19760af38/F1.jpg", "media/82b7cf0b81e19760af38/F2.jpg", "media/82b7cf0b81e19760af38/F3.jpg", "media/82b7cf0b81e19760af38/F4.jpg", "media/82b7cf0b81e19760af38/F5.jpg", "media/82b7cf0b81e19760af38/F6.jpg", "media/82b7cf0b81e19760af38/F7.jpg", "media/82b7cf0b81e19760af38/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (brown bull with white markings on its back, facing left) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64e127e6a351a2f51e3c4547adc7c0f59e775918bf8fb7918de4f4970d2596e3"}
{"qa_id": "MOSE__5968c2d8__camera_translation_direction__00", "video_id": "MOSE__5968c2d8", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "right", "forward"], "frame_paths": ["media/eec0ffcb77bd0fa37fd1/F0.jpg", "media/eec0ffcb77bd0fa37fd1/F1.jpg", "media/eec0ffcb77bd0fa37fd1/F2.jpg", "media/eec0ffcb77bd0fa37fd1/F3.jpg", "media/eec0ffcb77bd0fa37fd1/F4.jpg", "media/eec0ffcb77bd0fa37fd1/F5.jpg", "media/eec0ffcb77bd0fa37fd1/F6.jpg", "media/eec0ffcb77bd0fa37fd1/F7.jpg", "media/eec0ffcb77bd0fa37fd1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5e717415921c0ac412d25bf54fac8f1a3fdc0edc072fdcb9993260769dd1307"}
{"qa_id": "MOSE__59750be4__camera_translation_direction__00", "video_id": "MOSE__59750be4", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "left", "backward", "right"], "frame_paths": ["media/060ac0d9e53d9d8a6003/F0.jpg", "media/060ac0d9e53d9d8a6003/F1.jpg", "media/060ac0d9e53d9d8a6003/F2.jpg", "media/060ac0d9e53d9d8a6003/F3.jpg", "media/060ac0d9e53d9d8a6003/F4.jpg", "media/060ac0d9e53d9d8a6003/F5.jpg", "media/060ac0d9e53d9d8a6003/F6.jpg", "media/060ac0d9e53d9d8a6003/F7.jpg", "media/060ac0d9e53d9d8a6003/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d171f60b99d36e87f600ffc9f4d367b657f65235aa7465709292a842b17c85f"}
{"qa_id": "MOSE__59750be4__camera_view_direction_change__01", "video_id": "MOSE__59750be4", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/060ac0d9e53d9d8a6003/F0.jpg", "media/060ac0d9e53d9d8a6003/F1.jpg", "media/060ac0d9e53d9d8a6003/F2.jpg", "media/060ac0d9e53d9d8a6003/F3.jpg", "media/060ac0d9e53d9d8a6003/F4.jpg", "media/060ac0d9e53d9d8a6003/F5.jpg", "media/060ac0d9e53d9d8a6003/F6.jpg", "media/060ac0d9e53d9d8a6003/F7.jpg", "media/060ac0d9e53d9d8a6003/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "93626333f8ae00b8c5b7909c9540b956558a831446b9e15fc7ffbabc05439ca1"}
{"qa_id": "MOSE__5a723c30__camera_translation_direction__00", "video_id": "MOSE__5a723c30", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "down", "left"], "frame_paths": ["media/2ebe912877947b878784/F0.jpg", "media/2ebe912877947b878784/F1.jpg", "media/2ebe912877947b878784/F2.jpg", "media/2ebe912877947b878784/F3.jpg", "media/2ebe912877947b878784/F4.jpg", "media/2ebe912877947b878784/F5.jpg", "media/2ebe912877947b878784/F6.jpg", "media/2ebe912877947b878784/F7.jpg", "media/2ebe912877947b878784/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7a6894c6b496070eadf3a3074f73333f510f3945fd86bf834dce70ecfb119cb3"}
{"qa_id": "MOSE__5a723c30__object_world_motion_direction__01", "video_id": "MOSE__5a723c30", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (black horse with a light-colored halter) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/98c104f8c41a763068aa/F0.jpg", "media/98c104f8c41a763068aa/F1.jpg", "media/98c104f8c41a763068aa/F2.jpg", "media/98c104f8c41a763068aa/F3.jpg", "media/98c104f8c41a763068aa/F4.jpg", "media/98c104f8c41a763068aa/F5.jpg", "media/98c104f8c41a763068aa/F6.jpg", "media/98c104f8c41a763068aa/F7.jpg", "media/98c104f8c41a763068aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (black horse with a light-colored halter) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5b1d5dd83918e4de0fa1ae750ff6b6fa6b51bc6b3ded89a91ae49879de8d5a06"}
{"qa_id": "MOSE__5a92a0c1__camera_view_direction_change__00", "video_id": "MOSE__5a92a0c1", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/cd9d4cf308675eb9e226/F0.jpg", "media/cd9d4cf308675eb9e226/F1.jpg", "media/cd9d4cf308675eb9e226/F2.jpg", "media/cd9d4cf308675eb9e226/F3.jpg", "media/cd9d4cf308675eb9e226/F4.jpg", "media/cd9d4cf308675eb9e226/F5.jpg", "media/cd9d4cf308675eb9e226/F6.jpg", "media/cd9d4cf308675eb9e226/F7.jpg", "media/cd9d4cf308675eb9e226/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "MOSE__5a92a0c1__object_pair_distance_change__01", "video_id": "MOSE__5a92a0c1", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (black cow with white patches on the belly and legs) and object B (white cow with large black spots) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/b2980d0e36ce01d06d5f/F0.jpg", "media/b2980d0e36ce01d06d5f/F1.jpg", "media/b2980d0e36ce01d06d5f/F2.jpg", "media/b2980d0e36ce01d06d5f/F3.jpg", "media/b2980d0e36ce01d06d5f/F4.jpg", "media/b2980d0e36ce01d06d5f/F5.jpg", "media/b2980d0e36ce01d06d5f/F6.jpg", "media/b2980d0e36ce01d06d5f/F7.jpg", "media/b2980d0e36ce01d06d5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (black cow with white patches on the belly and legs) and object B (white cow with large black spots) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b244de61f6f68391c0a7c604a2d8903b054d3cc6d26f016f577283d00947bd1e"}
{"qa_id": "MOSE__5a92a0c1__object_travel_distance_comparison__02", "video_id": "MOSE__5a92a0c1", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black cow with white patches on the belly and legs) or B (white cow with large black spots)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/b2980d0e36ce01d06d5f/F0.jpg", "media/b2980d0e36ce01d06d5f/F1.jpg", "media/b2980d0e36ce01d06d5f/F2.jpg", "media/b2980d0e36ce01d06d5f/F3.jpg", "media/b2980d0e36ce01d06d5f/F4.jpg", "media/b2980d0e36ce01d06d5f/F5.jpg", "media/b2980d0e36ce01d06d5f/F6.jpg", "media/b2980d0e36ce01d06d5f/F7.jpg", "media/b2980d0e36ce01d06d5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black cow with white patches on the belly and legs) or B (white cow with large black spots)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f2a673e63fad5954f7f8644ccd781c7c596e1f67ea97d8e74769cfb8495ee66"}
{"qa_id": "MOSE__5ba84e3f__object_world_motion_direction__00", "video_id": "MOSE__5ba84e3f", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (puppy with light tan and white fur) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/65d32da16821b7716c4c/F0.jpg", "media/65d32da16821b7716c4c/F1.jpg", "media/65d32da16821b7716c4c/F2.jpg", "media/65d32da16821b7716c4c/F3.jpg", "media/65d32da16821b7716c4c/F4.jpg", "media/65d32da16821b7716c4c/F5.jpg", "media/65d32da16821b7716c4c/F6.jpg", "media/65d32da16821b7716c4c/F7.jpg", "media/65d32da16821b7716c4c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (puppy with light tan and white fur) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d4ef39c12d1e0ef7eb2a3686561b0ab12fdf33c686cf9898b7888b2abebfb6d1"}
{"qa_id": "MOSE__5bc7ea74__object_world_motion_direction__00", "video_id": "MOSE__5bc7ea74", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/20831aae65ce8d28bc40/F0.jpg", "media/20831aae65ce8d28bc40/F1.jpg", "media/20831aae65ce8d28bc40/F2.jpg", "media/20831aae65ce8d28bc40/F3.jpg", "media/20831aae65ce8d28bc40/F4.jpg", "media/20831aae65ce8d28bc40/F5.jpg", "media/20831aae65ce8d28bc40/F6.jpg", "media/20831aae65ce8d28bc40/F7.jpg", "media/20831aae65ce8d28bc40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d3d1e8cfa74bb2fccf939e8b47db54d9a5f148173a2e45814ec22b08cda79c34"}
{"qa_id": "MOSE__5bc7ea74__camera_object_distance_change__01", "video_id": "MOSE__5bc7ea74", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object A (white sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/20831aae65ce8d28bc40/F0.jpg", "media/20831aae65ce8d28bc40/F1.jpg", "media/20831aae65ce8d28bc40/F2.jpg", "media/20831aae65ce8d28bc40/F3.jpg", "media/20831aae65ce8d28bc40/F4.jpg", "media/20831aae65ce8d28bc40/F5.jpg", "media/20831aae65ce8d28bc40/F6.jpg", "media/20831aae65ce8d28bc40/F7.jpg", "media/20831aae65ce8d28bc40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (white sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "70711765de90ef4c654c662485f685467ad70764dcc2cb8652e785a856be3b28"}
{"qa_id": "MOSE__5bc7ea74__object_world_motion_direction__02", "video_id": "MOSE__5bc7ea74", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (green city bus) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "up", "backward", "forward"], "frame_paths": ["media/b7973b512ae028551def/F0.jpg", "media/b7973b512ae028551def/F1.jpg", "media/b7973b512ae028551def/F2.jpg", "media/b7973b512ae028551def/F3.jpg", "media/b7973b512ae028551def/F4.jpg", "media/b7973b512ae028551def/F5.jpg", "media/b7973b512ae028551def/F6.jpg", "media/b7973b512ae028551def/F7.jpg", "media/b7973b512ae028551def/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (green city bus) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "207f0a6346c6b749f665a88ee4985af799c4c026bf8d6cbd965567587cab6aa0"}
{"qa_id": "MOSE__5bc7ea74__camera_object_distance_change__03", "video_id": "MOSE__5bc7ea74", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object B (green city bus) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b7973b512ae028551def/F0.jpg", "media/b7973b512ae028551def/F1.jpg", "media/b7973b512ae028551def/F2.jpg", "media/b7973b512ae028551def/F3.jpg", "media/b7973b512ae028551def/F4.jpg", "media/b7973b512ae028551def/F5.jpg", "media/b7973b512ae028551def/F6.jpg", "media/b7973b512ae028551def/F7.jpg", "media/b7973b512ae028551def/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object B (green city bus) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31a2a1b91859752ffe291fcc1fcdd257809c7d83e5e89c1a7825066466bcf966"}
{"qa_id": "MOSE__5bc7ea74__object_pair_distance_change__04", "video_id": "MOSE__5bc7ea74", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did the 3D distance between object A (white sedan) and object B (green city bus) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/b1cd5c960ea7d85ee218/F0.jpg", "media/b1cd5c960ea7d85ee218/F1.jpg", "media/b1cd5c960ea7d85ee218/F2.jpg", "media/b1cd5c960ea7d85ee218/F3.jpg", "media/b1cd5c960ea7d85ee218/F4.jpg", "media/b1cd5c960ea7d85ee218/F5.jpg", "media/b1cd5c960ea7d85ee218/F6.jpg", "media/b1cd5c960ea7d85ee218/F7.jpg", "media/b1cd5c960ea7d85ee218/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (white sedan) and object B (green city bus) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a296dd1977f30c0e9b50c9d3f70c2bdd20a4b4f3841f4b9df62d7d571596265"}
{"qa_id": "MOSE__5bc7ea74__object_travel_distance_comparison__05", "video_id": "MOSE__5bc7ea74", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white sedan) or B (green city bus)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/b1cd5c960ea7d85ee218/F0.jpg", "media/b1cd5c960ea7d85ee218/F1.jpg", "media/b1cd5c960ea7d85ee218/F2.jpg", "media/b1cd5c960ea7d85ee218/F3.jpg", "media/b1cd5c960ea7d85ee218/F4.jpg", "media/b1cd5c960ea7d85ee218/F5.jpg", "media/b1cd5c960ea7d85ee218/F6.jpg", "media/b1cd5c960ea7d85ee218/F7.jpg", "media/b1cd5c960ea7d85ee218/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white sedan) or B (green city bus)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ace8497663dc651e311128239ed3b1b1638d1748359ab63bd13a6eb3105e0194"}
{"qa_id": "MOSE__5c80c82c__object_world_motion_direction__00", "video_id": "MOSE__5c80c82c", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (dog with dark brindle coat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/6a277483d707ace2b578/F0.jpg", "media/6a277483d707ace2b578/F1.jpg", "media/6a277483d707ace2b578/F2.jpg", "media/6a277483d707ace2b578/F3.jpg", "media/6a277483d707ace2b578/F4.jpg", "media/6a277483d707ace2b578/F5.jpg", "media/6a277483d707ace2b578/F6.jpg", "media/6a277483d707ace2b578/F7.jpg", "media/6a277483d707ace2b578/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (dog with dark brindle coat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3828cd08995bddf8c69d9629b6cafcaedcc326cf12757c64ebbd57a65b374da6"}
{"qa_id": "MOSE__5d2316f6__object_world_motion_direction__00", "video_id": "MOSE__5d2316f6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (brown goat grazing on the slope) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/f38822eee82803e7ae28/F0.jpg", "media/f38822eee82803e7ae28/F1.jpg", "media/f38822eee82803e7ae28/F2.jpg", "media/f38822eee82803e7ae28/F3.jpg", "media/f38822eee82803e7ae28/F4.jpg", "media/f38822eee82803e7ae28/F5.jpg", "media/f38822eee82803e7ae28/F6.jpg", "media/f38822eee82803e7ae28/F7.jpg", "media/f38822eee82803e7ae28/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (brown goat grazing on the slope) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d44d263252b4fd8c92d5261e5ceed57dc680bbb543fcfdbb10d31df49642f1e2"}
{"qa_id": "MOSE__5d2316f6__object_pair_distance_change__01", "video_id": "MOSE__5d2316f6", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (brown goat with a black face and horns) and object B (brown goat grazing on the slope) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/1fcf39a30902aa5072cf/F0.jpg", "media/1fcf39a30902aa5072cf/F1.jpg", "media/1fcf39a30902aa5072cf/F2.jpg", "media/1fcf39a30902aa5072cf/F3.jpg", "media/1fcf39a30902aa5072cf/F4.jpg", "media/1fcf39a30902aa5072cf/F5.jpg", "media/1fcf39a30902aa5072cf/F6.jpg", "media/1fcf39a30902aa5072cf/F7.jpg", "media/1fcf39a30902aa5072cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (brown goat with a black face and horns) and object B (brown goat grazing on the slope) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96c4fe73671b09d0aaadcaf5b09c3bb174ed2fa37af9b3d49388af7726b65924"}
{"qa_id": "MOSE__5df8490d__object_world_motion_direction__00", "video_id": "MOSE__5df8490d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (small baby elephant walking on the road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "backward", "left", "down"], "frame_paths": ["media/7405c5cf19be083aef81/F0.jpg", "media/7405c5cf19be083aef81/F1.jpg", "media/7405c5cf19be083aef81/F2.jpg", "media/7405c5cf19be083aef81/F3.jpg", "media/7405c5cf19be083aef81/F4.jpg", "media/7405c5cf19be083aef81/F5.jpg", "media/7405c5cf19be083aef81/F6.jpg", "media/7405c5cf19be083aef81/F7.jpg", "media/7405c5cf19be083aef81/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (small baby elephant walking on the road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4015d334f41f0f971411198eb41e4a38f0dcc538790688585ab670c58937c8fa"}
{"qa_id": "MOSE__5e167efa__camera_translation_direction__00", "video_id": "MOSE__5e167efa", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "up", "backward", "forward"], "frame_paths": ["media/b1babadfc10b2e2eecf6/F0.jpg", "media/b1babadfc10b2e2eecf6/F1.jpg", "media/b1babadfc10b2e2eecf6/F2.jpg", "media/b1babadfc10b2e2eecf6/F3.jpg", "media/b1babadfc10b2e2eecf6/F4.jpg", "media/b1babadfc10b2e2eecf6/F5.jpg", "media/b1babadfc10b2e2eecf6/F6.jpg", "media/b1babadfc10b2e2eecf6/F7.jpg", "media/b1babadfc10b2e2eecf6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "80ca8378f54cef7de51b3cd39b1967e682eab2d5d1be9c9a27e8aafb4ee2b87b"}
{"qa_id": "MOSE__5e167efa__object_pair_distance_change__02", "video_id": "MOSE__5e167efa", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did the 3D distance between object A (yellow fish) and object B (white fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/54955e9b4ce59e384f86/F0.jpg", "media/54955e9b4ce59e384f86/F1.jpg", "media/54955e9b4ce59e384f86/F2.jpg", "media/54955e9b4ce59e384f86/F3.jpg", "media/54955e9b4ce59e384f86/F4.jpg", "media/54955e9b4ce59e384f86/F5.jpg", "media/54955e9b4ce59e384f86/F6.jpg", "media/54955e9b4ce59e384f86/F7.jpg", "media/54955e9b4ce59e384f86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (yellow fish) and object B (white fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5c0bff208ba2f25b48e485bdfb1341b964cd02577f2ce85c4cd4f6781fe44a9"}
{"qa_id": "MOSE__5f3fcff6__object_world_motion_direction__00", "video_id": "MOSE__5f3fcff6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "forward", "left", "up"], "frame_paths": ["media/dafc95a7a699cee38a2c/F0.jpg", "media/dafc95a7a699cee38a2c/F1.jpg", "media/dafc95a7a699cee38a2c/F2.jpg", "media/dafc95a7a699cee38a2c/F3.jpg", "media/dafc95a7a699cee38a2c/F4.jpg", "media/dafc95a7a699cee38a2c/F5.jpg", "media/dafc95a7a699cee38a2c/F6.jpg", "media/dafc95a7a699cee38a2c/F7.jpg", "media/dafc95a7a699cee38a2c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. forward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0ff44559f231e8b5e7988dac81dd037c4d7f34da99561a56c7a5444659607f96"}
{"qa_id": "MOSE__5f3fcff6__object_world_motion_direction__01", "video_id": "MOSE__5f3fcff6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (smaller elephant walking behind the large adult) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "right", "up", "left"], "frame_paths": ["media/6ad1140de723e3e86220/F0.jpg", "media/6ad1140de723e3e86220/F1.jpg", "media/6ad1140de723e3e86220/F2.jpg", "media/6ad1140de723e3e86220/F3.jpg", "media/6ad1140de723e3e86220/F4.jpg", "media/6ad1140de723e3e86220/F5.jpg", "media/6ad1140de723e3e86220/F6.jpg", "media/6ad1140de723e3e86220/F7.jpg", "media/6ad1140de723e3e86220/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (smaller elephant walking behind the large adult) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e9cdda74033a6b87eb0a81959cffee91f23e5d2e227ac8904914cf680802c950"}
{"qa_id": "MOSE__5f54d077__object_world_motion_direction__00", "video_id": "MOSE__5f54d077", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (light brown chicken on the right side) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "right", "left"], "frame_paths": ["media/580efa21bb0c0450c832/F0.jpg", "media/580efa21bb0c0450c832/F1.jpg", "media/580efa21bb0c0450c832/F2.jpg", "media/580efa21bb0c0450c832/F3.jpg", "media/580efa21bb0c0450c832/F4.jpg", "media/580efa21bb0c0450c832/F5.jpg", "media/580efa21bb0c0450c832/F6.jpg", "media/580efa21bb0c0450c832/F7.jpg", "media/580efa21bb0c0450c832/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (light brown chicken on the right side) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ec66c72d6c5f98cb4743768a6ee9852af8ad4520258d14e7ede34bef4483f142"}
{"qa_id": "MOSE__5fbff256__object_pair_distance_change__00", "video_id": "MOSE__5fbff256", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did the 3D distance between object A (giraffe with long neck extending left) and object B (giraffe standing near the wooden fence) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/1ee5749fc3337889f74d/F0.jpg", "media/1ee5749fc3337889f74d/F1.jpg", "media/1ee5749fc3337889f74d/F2.jpg", "media/1ee5749fc3337889f74d/F3.jpg", "media/1ee5749fc3337889f74d/F4.jpg", "media/1ee5749fc3337889f74d/F5.jpg", "media/1ee5749fc3337889f74d/F6.jpg", "media/1ee5749fc3337889f74d/F7.jpg", "media/1ee5749fc3337889f74d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (giraffe with long neck extending left) and object B (giraffe standing near the wooden fence) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4b8ab23cddbe02d1ad24edd4a69d18cdc4bc51d14bcbb3d4e3bf136ee9eb5e4e"}
{"qa_id": "MOSE__5fbff256__object_travel_distance_comparison__01", "video_id": "MOSE__5fbff256", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (giraffe with long neck extending left) or B (giraffe standing near the wooden fence)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1ee5749fc3337889f74d/F0.jpg", "media/1ee5749fc3337889f74d/F1.jpg", "media/1ee5749fc3337889f74d/F2.jpg", "media/1ee5749fc3337889f74d/F3.jpg", "media/1ee5749fc3337889f74d/F4.jpg", "media/1ee5749fc3337889f74d/F5.jpg", "media/1ee5749fc3337889f74d/F6.jpg", "media/1ee5749fc3337889f74d/F7.jpg", "media/1ee5749fc3337889f74d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (giraffe with long neck extending left) or B (giraffe standing near the wooden fence)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65d930c9cae484e51a41bec8874b1a26806f2c9d6b63d4071048563c35a6057f"}
{"qa_id": "MOSE__5ff72128__object_world_motion_direction__00", "video_id": "MOSE__5ff72128", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in grey t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "left", "backward", "right"], "frame_paths": ["media/1097a0fc93a8a31b8757/F0.jpg", "media/1097a0fc93a8a31b8757/F1.jpg", "media/1097a0fc93a8a31b8757/F2.jpg", "media/1097a0fc93a8a31b8757/F3.jpg", "media/1097a0fc93a8a31b8757/F4.jpg", "media/1097a0fc93a8a31b8757/F5.jpg", "media/1097a0fc93a8a31b8757/F6.jpg", "media/1097a0fc93a8a31b8757/F7.jpg", "media/1097a0fc93a8a31b8757/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in grey t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f219d1acb718b554a3e349394e28e59f0244ca70d9069e089942f973bae86cf4"}
{"qa_id": "MOSE__5ff72128__camera_object_distance_change__01", "video_id": "MOSE__5ff72128", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (person in grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/1097a0fc93a8a31b8757/F0.jpg", "media/1097a0fc93a8a31b8757/F1.jpg", "media/1097a0fc93a8a31b8757/F2.jpg", "media/1097a0fc93a8a31b8757/F3.jpg", "media/1097a0fc93a8a31b8757/F4.jpg", "media/1097a0fc93a8a31b8757/F5.jpg", "media/1097a0fc93a8a31b8757/F6.jpg", "media/1097a0fc93a8a31b8757/F7.jpg", "media/1097a0fc93a8a31b8757/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (person in grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "130b82cdbccb0c2a8c567d7428b9a703aca1adb9565b6f38fa49ea693a1d0957"}
{"qa_id": "MOSE__5ff72128__object_pair_distance_change__02", "video_id": "MOSE__5ff72128", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (person in grey t-shirt) and object B (person in white t-shirt and black pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8d212fc1c46c459669af/F0.jpg", "media/8d212fc1c46c459669af/F1.jpg", "media/8d212fc1c46c459669af/F2.jpg", "media/8d212fc1c46c459669af/F3.jpg", "media/8d212fc1c46c459669af/F4.jpg", "media/8d212fc1c46c459669af/F5.jpg", "media/8d212fc1c46c459669af/F6.jpg", "media/8d212fc1c46c459669af/F7.jpg", "media/8d212fc1c46c459669af/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in grey t-shirt) and object B (person in white t-shirt and black pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "72abdbf0fc1510f7e744502a3d6cbe56a3ba95c1a4579afabf83534e20c51586"}
{"qa_id": "MOSE__60ab04c5__object_world_motion_direction__00", "video_id": "MOSE__60ab04c5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange shirt and blue jeans walking dogs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "down", "up", "forward"], "frame_paths": ["media/cfe299915fe77e3052e5/F0.jpg", "media/cfe299915fe77e3052e5/F1.jpg", "media/cfe299915fe77e3052e5/F2.jpg", "media/cfe299915fe77e3052e5/F3.jpg", "media/cfe299915fe77e3052e5/F4.jpg", "media/cfe299915fe77e3052e5/F5.jpg", "media/cfe299915fe77e3052e5/F6.jpg", "media/cfe299915fe77e3052e5/F7.jpg", "media/cfe299915fe77e3052e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange shirt and blue jeans walking dogs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e2eb52a99a338218b77f8a2ac0fe85ac178289fbfe6e1c5c54607a9ffc07ee7"}
{"qa_id": "MOSE__61a2e246__object_world_motion_direction__00", "video_id": "MOSE__61a2e246", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "forward", "down"], "frame_paths": ["media/dc587905646e6a85696a/F0.jpg", "media/dc587905646e6a85696a/F1.jpg", "media/dc587905646e6a85696a/F2.jpg", "media/dc587905646e6a85696a/F3.jpg", "media/dc587905646e6a85696a/F4.jpg", "media/dc587905646e6a85696a/F5.jpg", "media/dc587905646e6a85696a/F6.jpg", "media/dc587905646e6a85696a/F7.jpg", "media/dc587905646e6a85696a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "996dac84f4ea8705c98c39031201e2f809e7de9cb75741e12d17e520a4dbd3c0"}
{"qa_id": "MOSE__61cf0ec8__camera_object_distance_change__00", "video_id": "MOSE__61cf0ec8", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (white car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/40b03f09759590d1d74a/F0.jpg", "media/40b03f09759590d1d74a/F1.jpg", "media/40b03f09759590d1d74a/F2.jpg", "media/40b03f09759590d1d74a/F3.jpg", "media/40b03f09759590d1d74a/F4.jpg", "media/40b03f09759590d1d74a/F5.jpg", "media/40b03f09759590d1d74a/F6.jpg", "media/40b03f09759590d1d74a/F7.jpg", "media/40b03f09759590d1d74a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (white car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc9d97778260741455f5e9e6370148c4eadd7ab5d8bc043020803b77f76b728b"}
{"qa_id": "MOSE__61cf0ec8__camera_object_distance_change__01", "video_id": "MOSE__61cf0ec8", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object B (white car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/220dc7d9b31a9de32a8d/F0.jpg", "media/220dc7d9b31a9de32a8d/F1.jpg", "media/220dc7d9b31a9de32a8d/F2.jpg", "media/220dc7d9b31a9de32a8d/F3.jpg", "media/220dc7d9b31a9de32a8d/F4.jpg", "media/220dc7d9b31a9de32a8d/F5.jpg", "media/220dc7d9b31a9de32a8d/F6.jpg", "media/220dc7d9b31a9de32a8d/F7.jpg", "media/220dc7d9b31a9de32a8d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object B (white car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82dc81164f6d624cc85c6429935a5db32419441b3a237fabbcca6946651122d1"}
{"qa_id": "MOSE__61d80e22__object_world_motion_direction__00", "video_id": "MOSE__61d80e22", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (man in dark blue jacket and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/fdf690a9dfcfc2836202/F0.jpg", "media/fdf690a9dfcfc2836202/F1.jpg", "media/fdf690a9dfcfc2836202/F2.jpg", "media/fdf690a9dfcfc2836202/F3.jpg", "media/fdf690a9dfcfc2836202/F4.jpg", "media/fdf690a9dfcfc2836202/F5.jpg", "media/fdf690a9dfcfc2836202/F6.jpg", "media/fdf690a9dfcfc2836202/F7.jpg", "media/fdf690a9dfcfc2836202/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (man in dark blue jacket and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3936c45afdb7676d21f21b5b6912ee7a55bdcc0d7aa029852e00dd5e40232c41"}
{"qa_id": "MOSE__61d80e22__camera_object_distance_change__01", "video_id": "MOSE__61d80e22", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object A (man in dark blue jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fdf690a9dfcfc2836202/F0.jpg", "media/fdf690a9dfcfc2836202/F1.jpg", "media/fdf690a9dfcfc2836202/F2.jpg", "media/fdf690a9dfcfc2836202/F3.jpg", "media/fdf690a9dfcfc2836202/F4.jpg", "media/fdf690a9dfcfc2836202/F5.jpg", "media/fdf690a9dfcfc2836202/F6.jpg", "media/fdf690a9dfcfc2836202/F7.jpg", "media/fdf690a9dfcfc2836202/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (man in dark blue jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9cd861bb79608079943eb97d67e52ef93b35af37c6fc544775d253ab8f573b0b"}
{"qa_id": "MOSE__62132822__camera_translation_direction__00", "video_id": "MOSE__62132822", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "up", "down", "backward"], "frame_paths": ["media/8dd17c6c6465af230cb2/F0.jpg", "media/8dd17c6c6465af230cb2/F1.jpg", "media/8dd17c6c6465af230cb2/F2.jpg", "media/8dd17c6c6465af230cb2/F3.jpg", "media/8dd17c6c6465af230cb2/F4.jpg", "media/8dd17c6c6465af230cb2/F5.jpg", "media/8dd17c6c6465af230cb2/F6.jpg", "media/8dd17c6c6465af230cb2/F7.jpg", "media/8dd17c6c6465af230cb2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "882657d249d6a06e99b3f9bdbb5d7bbc096e0498f3fd7686256968cc5ffa0cf6"}
{"qa_id": "MOSE__631b82bd__camera_object_distance_change__00", "video_id": "MOSE__631b82bd", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/733957974e16af20a48f/F0.jpg", "media/733957974e16af20a48f/F1.jpg", "media/733957974e16af20a48f/F2.jpg", "media/733957974e16af20a48f/F3.jpg", "media/733957974e16af20a48f/F4.jpg", "media/733957974e16af20a48f/F5.jpg", "media/733957974e16af20a48f/F6.jpg", "media/733957974e16af20a48f/F7.jpg", "media/733957974e16af20a48f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48808ae17b7a6cea8c7f0924beecbc502ba163371f65c593a83126c973433268"}
{"qa_id": "MOSE__631b82bd__object_travel_distance_comparison__01", "video_id": "MOSE__631b82bd", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white SUV) or B (white SUV)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2ad1969fa09a18038420/F0.jpg", "media/2ad1969fa09a18038420/F1.jpg", "media/2ad1969fa09a18038420/F2.jpg", "media/2ad1969fa09a18038420/F3.jpg", "media/2ad1969fa09a18038420/F4.jpg", "media/2ad1969fa09a18038420/F5.jpg", "media/2ad1969fa09a18038420/F6.jpg", "media/2ad1969fa09a18038420/F7.jpg", "media/2ad1969fa09a18038420/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white SUV) or B (white SUV)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "75949d104d98059838e3c7d42fe6158e5e2e98b5a193e20ba2d99d949735a520"}
{"qa_id": "MOSE__63f138de__object_world_motion_direction__00", "video_id": "MOSE__63f138de", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (zebra with black and white stripes) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/65c733472293d0f1a89e/F0.jpg", "media/65c733472293d0f1a89e/F1.jpg", "media/65c733472293d0f1a89e/F2.jpg", "media/65c733472293d0f1a89e/F3.jpg", "media/65c733472293d0f1a89e/F4.jpg", "media/65c733472293d0f1a89e/F5.jpg", "media/65c733472293d0f1a89e/F6.jpg", "media/65c733472293d0f1a89e/F7.jpg", "media/65c733472293d0f1a89e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (zebra with black and white stripes) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "99cf96b57c4c9450de94e473cf0d94f34d968d518dacc93a8978290afdb9c85f"}
{"qa_id": "MOSE__63f138de__object_pair_distance_change__01", "video_id": "MOSE__63f138de", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did the 3D distance between object A (zebra with black and white stripes) and object B (zebra with black and white stripes) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/eb1f5d4a520655d999f2/F0.jpg", "media/eb1f5d4a520655d999f2/F1.jpg", "media/eb1f5d4a520655d999f2/F2.jpg", "media/eb1f5d4a520655d999f2/F3.jpg", "media/eb1f5d4a520655d999f2/F4.jpg", "media/eb1f5d4a520655d999f2/F5.jpg", "media/eb1f5d4a520655d999f2/F6.jpg", "media/eb1f5d4a520655d999f2/F7.jpg", "media/eb1f5d4a520655d999f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (zebra with black and white stripes) and object B (zebra with black and white stripes) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d99c0b26bacb8af00a75c811bb70f1bf09a1dc1020daeeedf7071b3f3f4f94ae"}
{"qa_id": "MOSE__6487110a__camera_translation_direction__00", "video_id": "MOSE__6487110a", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "backward", "down", "forward", "left"], "frame_paths": ["media/09a569875bac72c957e2/F0.jpg", "media/09a569875bac72c957e2/F1.jpg", "media/09a569875bac72c957e2/F2.jpg", "media/09a569875bac72c957e2/F3.jpg", "media/09a569875bac72c957e2/F4.jpg", "media/09a569875bac72c957e2/F5.jpg", "media/09a569875bac72c957e2/F6.jpg", "media/09a569875bac72c957e2/F7.jpg", "media/09a569875bac72c957e2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. backward\nD. down\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9e6d1d9c91dd591e9f0474b44e7feea2d8cb37d54af917a56e02674709c8b0c6"}
{"qa_id": "MOSE__64b89eb5__camera_object_distance_change__00", "video_id": "MOSE__64b89eb5", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (white van with black lower trim) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/796883ce2d52f7621174/F0.jpg", "media/796883ce2d52f7621174/F1.jpg", "media/796883ce2d52f7621174/F2.jpg", "media/796883ce2d52f7621174/F3.jpg", "media/796883ce2d52f7621174/F4.jpg", "media/796883ce2d52f7621174/F5.jpg", "media/796883ce2d52f7621174/F6.jpg", "media/796883ce2d52f7621174/F7.jpg", "media/796883ce2d52f7621174/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (white van with black lower trim) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe57a08bd4f2e004aff7fe782f61a30f6af6b644d01bd0c840771fefc1d50ada"}
{"qa_id": "MOSE__64b89eb5__object_pair_distance_change__01", "video_id": "MOSE__64b89eb5", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (white van with black lower trim) and object B (dark horse ridden by a person in dark clothing) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/f6ce8a65635a8a465580/F0.jpg", "media/f6ce8a65635a8a465580/F1.jpg", "media/f6ce8a65635a8a465580/F2.jpg", "media/f6ce8a65635a8a465580/F3.jpg", "media/f6ce8a65635a8a465580/F4.jpg", "media/f6ce8a65635a8a465580/F5.jpg", "media/f6ce8a65635a8a465580/F6.jpg", "media/f6ce8a65635a8a465580/F7.jpg", "media/f6ce8a65635a8a465580/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (white van with black lower trim) and object B (dark horse ridden by a person in dark clothing) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4822471a23a4a98a588dbf00f3317d455d4b0c6d8717877b599422547aa3879b"}
{"qa_id": "MOSE__64b89eb5__object_travel_distance_comparison__02", "video_id": "MOSE__64b89eb5", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white van with black lower trim) or B (dark horse ridden by a person in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f6ce8a65635a8a465580/F0.jpg", "media/f6ce8a65635a8a465580/F1.jpg", "media/f6ce8a65635a8a465580/F2.jpg", "media/f6ce8a65635a8a465580/F3.jpg", "media/f6ce8a65635a8a465580/F4.jpg", "media/f6ce8a65635a8a465580/F5.jpg", "media/f6ce8a65635a8a465580/F6.jpg", "media/f6ce8a65635a8a465580/F7.jpg", "media/f6ce8a65635a8a465580/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white van with black lower trim) or B (dark horse ridden by a person in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cac297da00a93a3cd97969d1ef3a43fc0d7c489c9ed77a4a74cec89b910d8a68"}
{"qa_id": "MOSE__657242e8__object_world_motion_direction__00", "video_id": "MOSE__657242e8", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (person with dark hair seen from behind) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "right", "left"], "frame_paths": ["media/0aeb36d78358d4135eb1/F0.jpg", "media/0aeb36d78358d4135eb1/F1.jpg", "media/0aeb36d78358d4135eb1/F2.jpg", "media/0aeb36d78358d4135eb1/F3.jpg", "media/0aeb36d78358d4135eb1/F4.jpg", "media/0aeb36d78358d4135eb1/F5.jpg", "media/0aeb36d78358d4135eb1/F6.jpg", "media/0aeb36d78358d4135eb1/F7.jpg", "media/0aeb36d78358d4135eb1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (person with dark hair seen from behind) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e70cba712dd236de62b5dd07c26a560ec1a7cf89d0fff88124d33f4e6a3d211"}
{"qa_id": "MOSE__657242e8__camera_object_distance_change__01", "video_id": "MOSE__657242e8", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, did object B (person with dark hair seen from behind) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/0aeb36d78358d4135eb1/F0.jpg", "media/0aeb36d78358d4135eb1/F1.jpg", "media/0aeb36d78358d4135eb1/F2.jpg", "media/0aeb36d78358d4135eb1/F3.jpg", "media/0aeb36d78358d4135eb1/F4.jpg", "media/0aeb36d78358d4135eb1/F5.jpg", "media/0aeb36d78358d4135eb1/F6.jpg", "media/0aeb36d78358d4135eb1/F7.jpg", "media/0aeb36d78358d4135eb1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object B (person with dark hair seen from behind) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da99f447b1066c7d39f74fce1228a84bf0795d2c15d1eee313cbe78ed58adf98"}
{"qa_id": "MOSE__65856fa0__camera_view_direction_change__00", "video_id": "MOSE__65856fa0", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/cf64d4376fcef79c87a9/F0.jpg", "media/cf64d4376fcef79c87a9/F1.jpg", "media/cf64d4376fcef79c87a9/F2.jpg", "media/cf64d4376fcef79c87a9/F3.jpg", "media/cf64d4376fcef79c87a9/F4.jpg", "media/cf64d4376fcef79c87a9/F5.jpg", "media/cf64d4376fcef79c87a9/F6.jpg", "media/cf64d4376fcef79c87a9/F7.jpg", "media/cf64d4376fcef79c87a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64e0fec494c815ab1788b70a6ff118230da5d701e043aa379365e44f1a46e68d"}
{"qa_id": "MOSE__65856fa0__object_world_motion_direction__01", "video_id": "MOSE__65856fa0", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (large adult elephant walking to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "right", "left"], "frame_paths": ["media/9756001a5699dbc3749e/F0.jpg", "media/9756001a5699dbc3749e/F1.jpg", "media/9756001a5699dbc3749e/F2.jpg", "media/9756001a5699dbc3749e/F3.jpg", "media/9756001a5699dbc3749e/F4.jpg", "media/9756001a5699dbc3749e/F5.jpg", "media/9756001a5699dbc3749e/F6.jpg", "media/9756001a5699dbc3749e/F7.jpg", "media/9756001a5699dbc3749e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (large adult elephant walking to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d482f57e42f800d378c0274e75854a23fe67922548437951ef5d06f9d7a4c0e"}
{"qa_id": "MOSE__65856fa0__camera_object_distance_change__02", "video_id": "MOSE__65856fa0", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did object A (large adult elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9756001a5699dbc3749e/F0.jpg", "media/9756001a5699dbc3749e/F1.jpg", "media/9756001a5699dbc3749e/F2.jpg", "media/9756001a5699dbc3749e/F3.jpg", "media/9756001a5699dbc3749e/F4.jpg", "media/9756001a5699dbc3749e/F5.jpg", "media/9756001a5699dbc3749e/F6.jpg", "media/9756001a5699dbc3749e/F7.jpg", "media/9756001a5699dbc3749e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (large adult elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78317d5f5ee158b8ab6eb2689a36a84dcb8a0a188cb828c9388b34baaac143c2"}
{"qa_id": "MOSE__65856fa0__object_world_motion_direction__03", "video_id": "MOSE__65856fa0", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (small juvenile elephant walking to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "backward", "forward"], "frame_paths": ["media/b8db006c44a3de2baa82/F0.jpg", "media/b8db006c44a3de2baa82/F1.jpg", "media/b8db006c44a3de2baa82/F2.jpg", "media/b8db006c44a3de2baa82/F3.jpg", "media/b8db006c44a3de2baa82/F4.jpg", "media/b8db006c44a3de2baa82/F5.jpg", "media/b8db006c44a3de2baa82/F6.jpg", "media/b8db006c44a3de2baa82/F7.jpg", "media/b8db006c44a3de2baa82/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (small juvenile elephant walking to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f33041fdde8e4d00353d3d89be830bbcddb5c28ed0b339472272157823db61d0"}
{"qa_id": "MOSE__65856fa0__camera_object_distance_change__04", "video_id": "MOSE__65856fa0", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did object B (small juvenile elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b8db006c44a3de2baa82/F0.jpg", "media/b8db006c44a3de2baa82/F1.jpg", "media/b8db006c44a3de2baa82/F2.jpg", "media/b8db006c44a3de2baa82/F3.jpg", "media/b8db006c44a3de2baa82/F4.jpg", "media/b8db006c44a3de2baa82/F5.jpg", "media/b8db006c44a3de2baa82/F6.jpg", "media/b8db006c44a3de2baa82/F7.jpg", "media/b8db006c44a3de2baa82/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object B (small juvenile elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "721e1635715aaa7609f9e112c5b76be0f7e0f57a88b34590f208b2ac07ad6763"}
{"qa_id": "MOSE__65856fa0__object_pair_distance_change__05", "video_id": "MOSE__65856fa0", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did the 3D distance between object A (large adult elephant walking to the right) and object B (small juvenile elephant walking to the right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4fe1c64e156487e09171/F0.jpg", "media/4fe1c64e156487e09171/F1.jpg", "media/4fe1c64e156487e09171/F2.jpg", "media/4fe1c64e156487e09171/F3.jpg", "media/4fe1c64e156487e09171/F4.jpg", "media/4fe1c64e156487e09171/F5.jpg", "media/4fe1c64e156487e09171/F6.jpg", "media/4fe1c64e156487e09171/F7.jpg", "media/4fe1c64e156487e09171/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (large adult elephant walking to the right) and object B (small juvenile elephant walking to the right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "33abc262d4b122891c1d177220f5968061e8951b20960f0fe01ed17bbe2ac978"}
{"qa_id": "MOSE__65856fa0__object_travel_distance_comparison__06", "video_id": "MOSE__65856fa0", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking to the right) or B (small juvenile elephant walking to the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4fe1c64e156487e09171/F0.jpg", "media/4fe1c64e156487e09171/F1.jpg", "media/4fe1c64e156487e09171/F2.jpg", "media/4fe1c64e156487e09171/F3.jpg", "media/4fe1c64e156487e09171/F4.jpg", "media/4fe1c64e156487e09171/F5.jpg", "media/4fe1c64e156487e09171/F6.jpg", "media/4fe1c64e156487e09171/F7.jpg", "media/4fe1c64e156487e09171/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking to the right) or B (small juvenile elephant walking to the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "479e958e7b0e60771d12fe4e5c9a09cf635b0d435aa7a4c67fb55c35e85b08f7"}
{"qa_id": "MOSE__67408899__object_world_motion_direction__00", "video_id": "MOSE__67408899", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (silver suv) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "forward", "left", "down", "right"], "frame_paths": ["media/dfba776d964429d80312/F0.jpg", "media/dfba776d964429d80312/F1.jpg", "media/dfba776d964429d80312/F2.jpg", "media/dfba776d964429d80312/F3.jpg", "media/dfba776d964429d80312/F4.jpg", "media/dfba776d964429d80312/F5.jpg", "media/dfba776d964429d80312/F6.jpg", "media/dfba776d964429d80312/F7.jpg", "media/dfba776d964429d80312/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (silver suv) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. forward\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "51cdbfbd8daa18e50b5129194c199deda816159caf350a3fc927c153d561c6ac"}
{"qa_id": "MOSE__67408899__camera_object_distance_change__01", "video_id": "MOSE__67408899", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F5, did object A (silver suv) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/dfba776d964429d80312/F0.jpg", "media/dfba776d964429d80312/F1.jpg", "media/dfba776d964429d80312/F2.jpg", "media/dfba776d964429d80312/F3.jpg", "media/dfba776d964429d80312/F4.jpg", "media/dfba776d964429d80312/F5.jpg", "media/dfba776d964429d80312/F6.jpg", "media/dfba776d964429d80312/F7.jpg", "media/dfba776d964429d80312/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (silver suv) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a9386151d078135d88505c27da2905bbff8b3352dcf3d14f3b59c9078f1493de"}
{"qa_id": "MOSE__6783a303__camera_object_distance_change__00", "video_id": "MOSE__6783a303", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (silver minivan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/abff0b9432ca562b138e/F0.jpg", "media/abff0b9432ca562b138e/F1.jpg", "media/abff0b9432ca562b138e/F2.jpg", "media/abff0b9432ca562b138e/F3.jpg", "media/abff0b9432ca562b138e/F4.jpg", "media/abff0b9432ca562b138e/F5.jpg", "media/abff0b9432ca562b138e/F6.jpg", "media/abff0b9432ca562b138e/F7.jpg", "media/abff0b9432ca562b138e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (silver minivan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0c580e52313959488760bb30be329cda797e9557b300edd547b772aba4c754f"}
{"qa_id": "MOSE__68954daf__object_world_motion_direction__00", "video_id": "MOSE__68954daf", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (large elephant walking on the dirt shoulder to the right of the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "backward", "up"], "frame_paths": ["media/756e108c9b1b4c7758cd/F0.jpg", "media/756e108c9b1b4c7758cd/F1.jpg", "media/756e108c9b1b4c7758cd/F2.jpg", "media/756e108c9b1b4c7758cd/F3.jpg", "media/756e108c9b1b4c7758cd/F4.jpg", "media/756e108c9b1b4c7758cd/F5.jpg", "media/756e108c9b1b4c7758cd/F6.jpg", "media/756e108c9b1b4c7758cd/F7.jpg", "media/756e108c9b1b4c7758cd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (large elephant walking on the dirt shoulder to the right of the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e18c86a6d6ed096f9a6d04eca95403beea0b1885a7553fb28af4eec781325ef"}
{"qa_id": "MOSE__68c879f2__camera_translation_direction__00", "video_id": "MOSE__68c879f2", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "up", "forward", "left"], "frame_paths": ["media/8d0d9f8b971eec9d5928/F0.jpg", "media/8d0d9f8b971eec9d5928/F1.jpg", "media/8d0d9f8b971eec9d5928/F2.jpg", "media/8d0d9f8b971eec9d5928/F3.jpg", "media/8d0d9f8b971eec9d5928/F4.jpg", "media/8d0d9f8b971eec9d5928/F5.jpg", "media/8d0d9f8b971eec9d5928/F6.jpg", "media/8d0d9f8b971eec9d5928/F7.jpg", "media/8d0d9f8b971eec9d5928/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1ff6eeaf9c5e1b5c5d767745e6ced9e90a02a45293e5f02ff1bcc15fe4be4330"}
{"qa_id": "MOSE__6a3bc149__object_world_motion_direction__00", "video_id": "MOSE__6a3bc149", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object B (sea turtle on the left with rear flippers visible) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "down", "up", "backward"], "frame_paths": ["media/7983a9442a536f40f28c/F0.jpg", "media/7983a9442a536f40f28c/F1.jpg", "media/7983a9442a536f40f28c/F2.jpg", "media/7983a9442a536f40f28c/F3.jpg", "media/7983a9442a536f40f28c/F4.jpg", "media/7983a9442a536f40f28c/F5.jpg", "media/7983a9442a536f40f28c/F6.jpg", "media/7983a9442a536f40f28c/F7.jpg", "media/7983a9442a536f40f28c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object B (sea turtle on the left with rear flippers visible) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ed1daad9f25054c1aef27bf960637a6420c3150a69f11704c1d0924f4b8ce01a"}
{"qa_id": "MOSE__6a3bc149__object_pair_distance_change__01", "video_id": "MOSE__6a3bc149", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did the 3D distance between object A (sea turtle on the right with front flippers raised) and object B (sea turtle on the left with rear flippers visible) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/722da8fc379c475f7c43/F0.jpg", "media/722da8fc379c475f7c43/F1.jpg", "media/722da8fc379c475f7c43/F2.jpg", "media/722da8fc379c475f7c43/F3.jpg", "media/722da8fc379c475f7c43/F4.jpg", "media/722da8fc379c475f7c43/F5.jpg", "media/722da8fc379c475f7c43/F6.jpg", "media/722da8fc379c475f7c43/F7.jpg", "media/722da8fc379c475f7c43/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (sea turtle on the right with front flippers raised) and object B (sea turtle on the left with rear flippers visible) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5845481d6565ce2712b8442cb0f70df13e3d0548ea60df24269efe87f0ae910c"}
{"qa_id": "MOSE__6a428397__object_world_motion_direction__00", "video_id": "MOSE__6a428397", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (brown horse standing near a wooden post) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "forward", "up", "right"], "frame_paths": ["media/49e117a96dd3ed3cc9c6/F0.jpg", "media/49e117a96dd3ed3cc9c6/F1.jpg", "media/49e117a96dd3ed3cc9c6/F2.jpg", "media/49e117a96dd3ed3cc9c6/F3.jpg", "media/49e117a96dd3ed3cc9c6/F4.jpg", "media/49e117a96dd3ed3cc9c6/F5.jpg", "media/49e117a96dd3ed3cc9c6/F6.jpg", "media/49e117a96dd3ed3cc9c6/F7.jpg", "media/49e117a96dd3ed3cc9c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (brown horse standing near a wooden post) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. forward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9656f6e3aeda3dbc09f975b325da0573d04bac9a6d95dbca935f2882452f2e07"}
{"qa_id": "MOSE__6a6e0e3b__object_pair_distance_change__00", "video_id": "MOSE__6a6e0e3b", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F4, did the 3D distance between object A (white cat sitting on the floor) and object B (grey tabby kitten walking on the floor) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/20b37558253bc0f7c232/F0.jpg", "media/20b37558253bc0f7c232/F1.jpg", "media/20b37558253bc0f7c232/F2.jpg", "media/20b37558253bc0f7c232/F3.jpg", "media/20b37558253bc0f7c232/F4.jpg", "media/20b37558253bc0f7c232/F5.jpg", "media/20b37558253bc0f7c232/F6.jpg", "media/20b37558253bc0f7c232/F7.jpg", "media/20b37558253bc0f7c232/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (white cat sitting on the floor) and object B (grey tabby kitten walking on the floor) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "67d8bad428b8cc344b93641d98c771e9b6f85fb1f0b661f85deaf43601670367"}
{"qa_id": "MOSE__6a6e0e3b__object_travel_distance_comparison__01", "video_id": "MOSE__6a6e0e3b", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white cat sitting on the floor) or B (grey tabby kitten walking on the floor)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/20b37558253bc0f7c232/F0.jpg", "media/20b37558253bc0f7c232/F1.jpg", "media/20b37558253bc0f7c232/F2.jpg", "media/20b37558253bc0f7c232/F3.jpg", "media/20b37558253bc0f7c232/F4.jpg", "media/20b37558253bc0f7c232/F5.jpg", "media/20b37558253bc0f7c232/F6.jpg", "media/20b37558253bc0f7c232/F7.jpg", "media/20b37558253bc0f7c232/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white cat sitting on the floor) or B (grey tabby kitten walking on the floor)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2a4f77dad85dd6d8bb5d7d2f9685dc75a820b5b9c661f0300fd8204a3c57b07"}
{"qa_id": "MOSE__6adad2f7__object_world_motion_direction__00", "video_id": "MOSE__6adad2f7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (white cat with orange markings) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "down", "forward"], "frame_paths": ["media/7d1229631bbabf902773/F0.jpg", "media/7d1229631bbabf902773/F1.jpg", "media/7d1229631bbabf902773/F2.jpg", "media/7d1229631bbabf902773/F3.jpg", "media/7d1229631bbabf902773/F4.jpg", "media/7d1229631bbabf902773/F5.jpg", "media/7d1229631bbabf902773/F6.jpg", "media/7d1229631bbabf902773/F7.jpg", "media/7d1229631bbabf902773/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (white cat with orange markings) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "441553791a1a9c9d38fe0757c7dafa978cefdaf395b9b336d4bed6806d4a4acc"}
{"qa_id": "MOSE__6adad2f7__camera_object_distance_change__01", "video_id": "MOSE__6adad2f7", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (white cat with orange markings) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7d1229631bbabf902773/F0.jpg", "media/7d1229631bbabf902773/F1.jpg", "media/7d1229631bbabf902773/F2.jpg", "media/7d1229631bbabf902773/F3.jpg", "media/7d1229631bbabf902773/F4.jpg", "media/7d1229631bbabf902773/F5.jpg", "media/7d1229631bbabf902773/F6.jpg", "media/7d1229631bbabf902773/F7.jpg", "media/7d1229631bbabf902773/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (white cat with orange markings) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "195031b62f3c790083a9d5808aa99a0efc03d247cf10dccee1b21fd2c8a3e173"}
{"qa_id": "MOSE__6adad2f7__object_travel_distance_comparison__02", "video_id": "MOSE__6adad2f7", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white cat with orange markings) or B (small dark kitten)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4ffa943739e8bf5f9460/F0.jpg", "media/4ffa943739e8bf5f9460/F1.jpg", "media/4ffa943739e8bf5f9460/F2.jpg", "media/4ffa943739e8bf5f9460/F3.jpg", "media/4ffa943739e8bf5f9460/F4.jpg", "media/4ffa943739e8bf5f9460/F5.jpg", "media/4ffa943739e8bf5f9460/F6.jpg", "media/4ffa943739e8bf5f9460/F7.jpg", "media/4ffa943739e8bf5f9460/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white cat with orange markings) or B (small dark kitten)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "158243c9bcc0ade728f37d7f60f054a8c50aba3285666308c93374d0161c0301"}
{"qa_id": "MOSE__6da17988__object_pair_distance_change__00", "video_id": "MOSE__6da17988", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (large adult elephant standing in the group) and object B (young elephant walking away from the camera) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/d5e3d31254da831efdcf/F0.jpg", "media/d5e3d31254da831efdcf/F1.jpg", "media/d5e3d31254da831efdcf/F2.jpg", "media/d5e3d31254da831efdcf/F3.jpg", "media/d5e3d31254da831efdcf/F4.jpg", "media/d5e3d31254da831efdcf/F5.jpg", "media/d5e3d31254da831efdcf/F6.jpg", "media/d5e3d31254da831efdcf/F7.jpg", "media/d5e3d31254da831efdcf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g10"}, "requested_object_ids": ["g2", "g10"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (large adult elephant standing in the group) and object B (young elephant walking away from the camera) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a2e3b6f610f3c7c2475342464072d5909ce74dc3cbd2895884267ec0051607c0"}
{"qa_id": "MOSE__6df96f15__camera_object_distance_change__00", "video_id": "MOSE__6df96f15", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did object A (orange fish) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8f698a83be62312d8d49/F0.jpg", "media/8f698a83be62312d8d49/F1.jpg", "media/8f698a83be62312d8d49/F2.jpg", "media/8f698a83be62312d8d49/F3.jpg", "media/8f698a83be62312d8d49/F4.jpg", "media/8f698a83be62312d8d49/F5.jpg", "media/8f698a83be62312d8d49/F6.jpg", "media/8f698a83be62312d8d49/F7.jpg", "media/8f698a83be62312d8d49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object A (orange fish) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4c8356f97db3108e1b5cd01da79f5f00a66abbb7a36dd683a15896fd98afbb50"}
{"qa_id": "MOSE__6df96f15__object_pair_distance_change__01", "video_id": "MOSE__6df96f15", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (orange fish) and object B (pinkish-orange fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/a27841860291ecf0ee2b/F0.jpg", "media/a27841860291ecf0ee2b/F1.jpg", "media/a27841860291ecf0ee2b/F2.jpg", "media/a27841860291ecf0ee2b/F3.jpg", "media/a27841860291ecf0ee2b/F4.jpg", "media/a27841860291ecf0ee2b/F5.jpg", "media/a27841860291ecf0ee2b/F6.jpg", "media/a27841860291ecf0ee2b/F7.jpg", "media/a27841860291ecf0ee2b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (orange fish) and object B (pinkish-orange fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cff31da5af4dce31e06a30e022cb153e3f51c2c9162b4246b464fa2e7442bca5"}
{"qa_id": "MOSE__6df96f15__object_travel_distance_comparison__02", "video_id": "MOSE__6df96f15", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish) or B (pinkish-orange fish)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/a27841860291ecf0ee2b/F0.jpg", "media/a27841860291ecf0ee2b/F1.jpg", "media/a27841860291ecf0ee2b/F2.jpg", "media/a27841860291ecf0ee2b/F3.jpg", "media/a27841860291ecf0ee2b/F4.jpg", "media/a27841860291ecf0ee2b/F5.jpg", "media/a27841860291ecf0ee2b/F6.jpg", "media/a27841860291ecf0ee2b/F7.jpg", "media/a27841860291ecf0ee2b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish) or B (pinkish-orange fish)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b63fd454a90a06769af8447c8d4486eebfaf17b35b1c5c7f62635ec65e34d4f2"}
{"qa_id": "MOSE__6e2dce1b__object_pair_distance_change__00", "video_id": "MOSE__6e2dce1b", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did the 3D distance between object A (adult giraffe with brown spots) and object B (smaller giraffe standing to the right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/a2df10d05436c064f802/F0.jpg", "media/a2df10d05436c064f802/F1.jpg", "media/a2df10d05436c064f802/F2.jpg", "media/a2df10d05436c064f802/F3.jpg", "media/a2df10d05436c064f802/F4.jpg", "media/a2df10d05436c064f802/F5.jpg", "media/a2df10d05436c064f802/F6.jpg", "media/a2df10d05436c064f802/F7.jpg", "media/a2df10d05436c064f802/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (adult giraffe with brown spots) and object B (smaller giraffe standing to the right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "678ed053c868f1589eabb7e2867b6d2ed6a6f3d604f14bfa2172dfc78aa82c71"}
{"qa_id": "MOSE__6f6e6c37__object_world_motion_direction__00", "video_id": "MOSE__6f6e6c37", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (fish with vertical black and white stripes) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "down", "left", "forward"], "frame_paths": ["media/9ddc79500a0598a43bcb/F0.jpg", "media/9ddc79500a0598a43bcb/F1.jpg", "media/9ddc79500a0598a43bcb/F2.jpg", "media/9ddc79500a0598a43bcb/F3.jpg", "media/9ddc79500a0598a43bcb/F4.jpg", "media/9ddc79500a0598a43bcb/F5.jpg", "media/9ddc79500a0598a43bcb/F6.jpg", "media/9ddc79500a0598a43bcb/F7.jpg", "media/9ddc79500a0598a43bcb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g8"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (fish with vertical black and white stripes) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "076240c404998a648ac1b7f7fb4b1d651b15524e282d3979e61aefde99aa64bb"}
{"qa_id": "MOSE__7040385d__camera_view_direction_change__00", "video_id": "MOSE__7040385d", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/21fa2263a1b11ec90871/F0.jpg", "media/21fa2263a1b11ec90871/F1.jpg", "media/21fa2263a1b11ec90871/F2.jpg", "media/21fa2263a1b11ec90871/F3.jpg", "media/21fa2263a1b11ec90871/F4.jpg", "media/21fa2263a1b11ec90871/F5.jpg", "media/21fa2263a1b11ec90871/F6.jpg", "media/21fa2263a1b11ec90871/F7.jpg", "media/21fa2263a1b11ec90871/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "MOSE__7040385d__object_world_motion_direction__01", "video_id": "MOSE__7040385d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (black motorcycle with a rider in a dark blue shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "forward", "backward", "left", "right"], "frame_paths": ["media/77e038c6fb26046c711a/F0.jpg", "media/77e038c6fb26046c711a/F1.jpg", "media/77e038c6fb26046c711a/F2.jpg", "media/77e038c6fb26046c711a/F3.jpg", "media/77e038c6fb26046c711a/F4.jpg", "media/77e038c6fb26046c711a/F5.jpg", "media/77e038c6fb26046c711a/F6.jpg", "media/77e038c6fb26046c711a/F7.jpg", "media/77e038c6fb26046c711a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (black motorcycle with a rider in a dark blue shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. forward\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2257e01b7dd061f0004a09a0da6230abe61c047d3ff956aeb46369fbd62c1f5f"}
{"qa_id": "MOSE__7040385d__camera_object_distance_change__02", "video_id": "MOSE__7040385d", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (black motorcycle with a rider in a dark blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/77e038c6fb26046c711a/F0.jpg", "media/77e038c6fb26046c711a/F1.jpg", "media/77e038c6fb26046c711a/F2.jpg", "media/77e038c6fb26046c711a/F3.jpg", "media/77e038c6fb26046c711a/F4.jpg", "media/77e038c6fb26046c711a/F5.jpg", "media/77e038c6fb26046c711a/F6.jpg", "media/77e038c6fb26046c711a/F7.jpg", "media/77e038c6fb26046c711a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (black motorcycle with a rider in a dark blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "27d3d99cf38f9f348a9e94c2d29003ea24da1cbe8c79be1a359a814a443af18d"}
{"qa_id": "MOSE__7040385d__object_world_motion_direction__03", "video_id": "MOSE__7040385d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object B (black motorcycle with a rider in a light blue shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "right", "up", "backward"], "frame_paths": ["media/3c59239fd7e95e109ca0/F0.jpg", "media/3c59239fd7e95e109ca0/F1.jpg", "media/3c59239fd7e95e109ca0/F2.jpg", "media/3c59239fd7e95e109ca0/F3.jpg", "media/3c59239fd7e95e109ca0/F4.jpg", "media/3c59239fd7e95e109ca0/F5.jpg", "media/3c59239fd7e95e109ca0/F6.jpg", "media/3c59239fd7e95e109ca0/F7.jpg", "media/3c59239fd7e95e109ca0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object B (black motorcycle with a rider in a light blue shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "41a02ab2dc7592b9fd44844d928ff5954ba1147838e3d59512889c8a6a7dbd0e"}
{"qa_id": "MOSE__7040385d__camera_object_distance_change__04", "video_id": "MOSE__7040385d", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object B (black motorcycle with a rider in a light blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/3c59239fd7e95e109ca0/F0.jpg", "media/3c59239fd7e95e109ca0/F1.jpg", "media/3c59239fd7e95e109ca0/F2.jpg", "media/3c59239fd7e95e109ca0/F3.jpg", "media/3c59239fd7e95e109ca0/F4.jpg", "media/3c59239fd7e95e109ca0/F5.jpg", "media/3c59239fd7e95e109ca0/F6.jpg", "media/3c59239fd7e95e109ca0/F7.jpg", "media/3c59239fd7e95e109ca0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object B (black motorcycle with a rider in a light blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8ff2d470d284c335a57eba923ef225c620f132bd7d8d9a02e88bb4acc4e62eb8"}
{"qa_id": "MOSE__71217dfc__object_world_motion_direction__00", "video_id": "MOSE__71217dfc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (white horse with a light mane) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "backward", "right", "up"], "frame_paths": ["media/2dde17e5a27836cfe849/F0.jpg", "media/2dde17e5a27836cfe849/F1.jpg", "media/2dde17e5a27836cfe849/F2.jpg", "media/2dde17e5a27836cfe849/F3.jpg", "media/2dde17e5a27836cfe849/F4.jpg", "media/2dde17e5a27836cfe849/F5.jpg", "media/2dde17e5a27836cfe849/F6.jpg", "media/2dde17e5a27836cfe849/F7.jpg", "media/2dde17e5a27836cfe849/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (white horse with a light mane) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f59cc5051c9620f7d475ba83278aca6ef51da2b1da619bd00a174c8ed44b3b69"}
{"qa_id": "MOSE__724a5d1c__object_world_motion_direction__00", "video_id": "MOSE__724a5d1c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (giant panda with black and white fur) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "right", "up", "forward", "left"], "frame_paths": ["media/d51f3ee521391da49ad9/F0.jpg", "media/d51f3ee521391da49ad9/F1.jpg", "media/d51f3ee521391da49ad9/F2.jpg", "media/d51f3ee521391da49ad9/F3.jpg", "media/d51f3ee521391da49ad9/F4.jpg", "media/d51f3ee521391da49ad9/F5.jpg", "media/d51f3ee521391da49ad9/F6.jpg", "media/d51f3ee521391da49ad9/F7.jpg", "media/d51f3ee521391da49ad9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (giant panda with black and white fur) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. right\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c59c71023f62bb9d6d3dc9e023c8c333bdf8cd538905b2d6494f294fd2fce938"}
{"qa_id": "MOSE__724cc810__object_world_motion_direction__00", "video_id": "MOSE__724cc810", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in orange life vest and blue helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/4bbff1962bbba97a9dfd/F0.jpg", "media/4bbff1962bbba97a9dfd/F1.jpg", "media/4bbff1962bbba97a9dfd/F2.jpg", "media/4bbff1962bbba97a9dfd/F3.jpg", "media/4bbff1962bbba97a9dfd/F4.jpg", "media/4bbff1962bbba97a9dfd/F5.jpg", "media/4bbff1962bbba97a9dfd/F6.jpg", "media/4bbff1962bbba97a9dfd/F7.jpg", "media/4bbff1962bbba97a9dfd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in orange life vest and blue helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3171f78eb64288ac13636ff0f9eb2b8698c9aedb492a1435365877ac56b53aab"}
{"qa_id": "MOSE__72b2b77d__camera_translation_direction__00", "video_id": "MOSE__72b2b77d", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/4ddbfff15eea1b800b32/F0.jpg", "media/4ddbfff15eea1b800b32/F1.jpg", "media/4ddbfff15eea1b800b32/F2.jpg", "media/4ddbfff15eea1b800b32/F3.jpg", "media/4ddbfff15eea1b800b32/F4.jpg", "media/4ddbfff15eea1b800b32/F5.jpg", "media/4ddbfff15eea1b800b32/F6.jpg", "media/4ddbfff15eea1b800b32/F7.jpg", "media/4ddbfff15eea1b800b32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84d72235cb5e1ffe9c608fb14bdbfe4c7e64cdc9fe86a9c032851002de63d6f1"}
{"qa_id": "MOSE__730bc486__object_world_motion_direction__00", "video_id": "MOSE__730bc486", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (black chicken with white markings on the wing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/6e2e84eb304d0e31a99a/F0.jpg", "media/6e2e84eb304d0e31a99a/F1.jpg", "media/6e2e84eb304d0e31a99a/F2.jpg", "media/6e2e84eb304d0e31a99a/F3.jpg", "media/6e2e84eb304d0e31a99a/F4.jpg", "media/6e2e84eb304d0e31a99a/F5.jpg", "media/6e2e84eb304d0e31a99a/F6.jpg", "media/6e2e84eb304d0e31a99a/F7.jpg", "media/6e2e84eb304d0e31a99a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g9"}, "requested_object_ids": ["g9"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (black chicken with white markings on the wing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7a2d1ef0b85e91cc42335576a0a57fff0d2a939d3c56ccac2569fd733acc48ab"}
{"qa_id": "MOSE__730bc486__object_pair_distance_change__01", "video_id": "MOSE__730bc486", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did the 3D distance between object A (grey and white goose with a long neck) and object B (black chicken with white markings on the wing) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/96519e574b26ebcc2f73/F0.jpg", "media/96519e574b26ebcc2f73/F1.jpg", "media/96519e574b26ebcc2f73/F2.jpg", "media/96519e574b26ebcc2f73/F3.jpg", "media/96519e574b26ebcc2f73/F4.jpg", "media/96519e574b26ebcc2f73/F5.jpg", "media/96519e574b26ebcc2f73/F6.jpg", "media/96519e574b26ebcc2f73/F7.jpg", "media/96519e574b26ebcc2f73/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g9"}, "requested_object_ids": ["g4", "g9"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (grey and white goose with a long neck) and object B (black chicken with white markings on the wing) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6363f17f3567660d894490dc55490a46cfddd170467bad4118a416bf0a1ae65c"}
{"qa_id": "MOSE__731149b3__camera_object_distance_change__00", "video_id": "MOSE__731149b3", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (red SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/70f3730b00fcb7889a34/F0.jpg", "media/70f3730b00fcb7889a34/F1.jpg", "media/70f3730b00fcb7889a34/F2.jpg", "media/70f3730b00fcb7889a34/F3.jpg", "media/70f3730b00fcb7889a34/F4.jpg", "media/70f3730b00fcb7889a34/F5.jpg", "media/70f3730b00fcb7889a34/F6.jpg", "media/70f3730b00fcb7889a34/F7.jpg", "media/70f3730b00fcb7889a34/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (red SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "60b474c57ff801d4223c957cf83b6c6fcaeef398a6f1c6a905c78f9f78f952ca"}
{"qa_id": "MOSE__74107936__object_travel_distance_comparison__00", "video_id": "MOSE__74107936", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (adult tiger with orange and black stripes walking in the center) or B (tiger with orange and black stripes walking to the left of the central tiger)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/f89f1213c52ba0870ea3/F0.jpg", "media/f89f1213c52ba0870ea3/F1.jpg", "media/f89f1213c52ba0870ea3/F2.jpg", "media/f89f1213c52ba0870ea3/F3.jpg", "media/f89f1213c52ba0870ea3/F4.jpg", "media/f89f1213c52ba0870ea3/F5.jpg", "media/f89f1213c52ba0870ea3/F6.jpg", "media/f89f1213c52ba0870ea3/F7.jpg", "media/f89f1213c52ba0870ea3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (adult tiger with orange and black stripes walking in the center) or B (tiger with orange and black stripes walking to the left of the central tiger)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "11462967f73af0fd990ffa0d060d8bf1a08e4c1a4eeff412f311c889fa23e135"}
{"qa_id": "MOSE__75d26197__camera_translation_direction__00", "video_id": "MOSE__75d26197", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "forward", "up", "backward", "right"], "frame_paths": ["media/6bfcbb166ab5e9faa595/F0.jpg", "media/6bfcbb166ab5e9faa595/F1.jpg", "media/6bfcbb166ab5e9faa595/F2.jpg", "media/6bfcbb166ab5e9faa595/F3.jpg", "media/6bfcbb166ab5e9faa595/F4.jpg", "media/6bfcbb166ab5e9faa595/F5.jpg", "media/6bfcbb166ab5e9faa595/F6.jpg", "media/6bfcbb166ab5e9faa595/F7.jpg", "media/6bfcbb166ab5e9faa595/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. forward\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9acaa006771a042fe8717b9f965dad22c7060d5475ed42aa1a10644e5681b8a1"}
{"qa_id": "MOSE__75d26197__object_travel_distance_comparison__01", "video_id": "MOSE__75d26197", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large turtle with a dark, patterned shell positioned on top) or B (turtle with a lighter, wrinkled head and neck positioned underneath)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/42b872fd275037783ac0/F0.jpg", "media/42b872fd275037783ac0/F1.jpg", "media/42b872fd275037783ac0/F2.jpg", "media/42b872fd275037783ac0/F3.jpg", "media/42b872fd275037783ac0/F4.jpg", "media/42b872fd275037783ac0/F5.jpg", "media/42b872fd275037783ac0/F6.jpg", "media/42b872fd275037783ac0/F7.jpg", "media/42b872fd275037783ac0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large turtle with a dark, patterned shell positioned on top) or B (turtle with a lighter, wrinkled head and neck positioned underneath)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "26e98e51f7a5377b4439e9584d1b39ef2a28bea645f15d58e2071ecdd74690e2"}
{"qa_id": "MOSE__77e246da__camera_translation_direction__00", "video_id": "MOSE__77e246da", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/415ef7e5bf89f242f2a8/F0.jpg", "media/415ef7e5bf89f242f2a8/F1.jpg", "media/415ef7e5bf89f242f2a8/F2.jpg", "media/415ef7e5bf89f242f2a8/F3.jpg", "media/415ef7e5bf89f242f2a8/F4.jpg", "media/415ef7e5bf89f242f2a8/F5.jpg", "media/415ef7e5bf89f242f2a8/F6.jpg", "media/415ef7e5bf89f242f2a8/F7.jpg", "media/415ef7e5bf89f242f2a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1311cf29b5f32900ea6aff62954de1b94a7c7fe7cc6c3a4fdb7e31542f7f3fc7"}
{"qa_id": "MOSE__792c6ff7__object_world_motion_direction__00", "video_id": "MOSE__792c6ff7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (fish with vertical black and white stripes) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "down", "up", "forward"], "frame_paths": ["media/a9a1a1c84253d01bc280/F0.jpg", "media/a9a1a1c84253d01bc280/F1.jpg", "media/a9a1a1c84253d01bc280/F2.jpg", "media/a9a1a1c84253d01bc280/F3.jpg", "media/a9a1a1c84253d01bc280/F4.jpg", "media/a9a1a1c84253d01bc280/F5.jpg", "media/a9a1a1c84253d01bc280/F6.jpg", "media/a9a1a1c84253d01bc280/F7.jpg", "media/a9a1a1c84253d01bc280/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (fish with vertical black and white stripes) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "573738db15ef97220ad14a06fa6372acbdd89acdd9b137e514c11e4c808b0351"}
{"qa_id": "MOSE__7981f344__object_world_motion_direction__00", "video_id": "MOSE__7981f344", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange coat) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "forward", "right", "left", "down"], "frame_paths": ["media/e830e0694a4cd0df4f2d/F0.jpg", "media/e830e0694a4cd0df4f2d/F1.jpg", "media/e830e0694a4cd0df4f2d/F2.jpg", "media/e830e0694a4cd0df4f2d/F3.jpg", "media/e830e0694a4cd0df4f2d/F4.jpg", "media/e830e0694a4cd0df4f2d/F5.jpg", "media/e830e0694a4cd0df4f2d/F6.jpg", "media/e830e0694a4cd0df4f2d/F7.jpg", "media/e830e0694a4cd0df4f2d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange coat) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5a2f4f8048e73dbc380085b93c46aaeeca5308da5a3f4513b4f882cb5b4f653a"}
{"qa_id": "MOSE__79af6097__object_world_motion_direction__00", "video_id": "MOSE__79af6097", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object B (white goose with black spots) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "right", "forward"], "frame_paths": ["media/d0db48e855140f0fcf34/F0.jpg", "media/d0db48e855140f0fcf34/F1.jpg", "media/d0db48e855140f0fcf34/F2.jpg", "media/d0db48e855140f0fcf34/F3.jpg", "media/d0db48e855140f0fcf34/F4.jpg", "media/d0db48e855140f0fcf34/F5.jpg", "media/d0db48e855140f0fcf34/F6.jpg", "media/d0db48e855140f0fcf34/F7.jpg", "media/d0db48e855140f0fcf34/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object B (white goose with black spots) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4ece75cbd8831b43c194385ed3627f337265b1d9df0df40adb42459e72aa9aa2"}
{"qa_id": "MOSE__79af6097__camera_object_distance_change__01", "video_id": "MOSE__79af6097", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did object B (white goose with black spots) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d0db48e855140f0fcf34/F0.jpg", "media/d0db48e855140f0fcf34/F1.jpg", "media/d0db48e855140f0fcf34/F2.jpg", "media/d0db48e855140f0fcf34/F3.jpg", "media/d0db48e855140f0fcf34/F4.jpg", "media/d0db48e855140f0fcf34/F5.jpg", "media/d0db48e855140f0fcf34/F6.jpg", "media/d0db48e855140f0fcf34/F7.jpg", "media/d0db48e855140f0fcf34/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object B (white goose with black spots) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a6a22e1b437f069fe8b2eeefc1c75cb387efa28ba07ab0a0810314fe32bdbdb"}
{"qa_id": "MOSE__79af6097__object_pair_distance_change__02", "video_id": "MOSE__79af6097", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did the 3D distance between object A (white goose with orange beak) and object B (white goose with black spots) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4fc6c72782290fff0378/F0.jpg", "media/4fc6c72782290fff0378/F1.jpg", "media/4fc6c72782290fff0378/F2.jpg", "media/4fc6c72782290fff0378/F3.jpg", "media/4fc6c72782290fff0378/F4.jpg", "media/4fc6c72782290fff0378/F5.jpg", "media/4fc6c72782290fff0378/F6.jpg", "media/4fc6c72782290fff0378/F7.jpg", "media/4fc6c72782290fff0378/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (white goose with orange beak) and object B (white goose with black spots) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "080836989ac53941a4f4879cc5c764da09a6a4361066890263a41f559f9470ed"}
{"qa_id": "MOSE__7ade3068__camera_view_direction_change__00", "video_id": "MOSE__7ade3068", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/4a19b3d3cf7fea43b28e/F0.jpg", "media/4a19b3d3cf7fea43b28e/F1.jpg", "media/4a19b3d3cf7fea43b28e/F2.jpg", "media/4a19b3d3cf7fea43b28e/F3.jpg", "media/4a19b3d3cf7fea43b28e/F4.jpg", "media/4a19b3d3cf7fea43b28e/F5.jpg", "media/4a19b3d3cf7fea43b28e/F6.jpg", "media/4a19b3d3cf7fea43b28e/F7.jpg", "media/4a19b3d3cf7fea43b28e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "be34eaa31a225674aefc50aee5da8d7bd0faad3ba6aa7bcedc9b5f8db21ac377"}
{"qa_id": "MOSE__7ade3068__object_world_motion_direction__01", "video_id": "MOSE__7ade3068", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object A (brown bull with white markings on its back) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "left", "right", "backward"], "frame_paths": ["media/631e31f3166faeef7668/F0.jpg", "media/631e31f3166faeef7668/F1.jpg", "media/631e31f3166faeef7668/F2.jpg", "media/631e31f3166faeef7668/F3.jpg", "media/631e31f3166faeef7668/F4.jpg", "media/631e31f3166faeef7668/F5.jpg", "media/631e31f3166faeef7668/F6.jpg", "media/631e31f3166faeef7668/F7.jpg", "media/631e31f3166faeef7668/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object A (brown bull with white markings on its back) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c205d205a847f33de04b4b15ee7907734e4fb7f828fe6a2a338108774c798ac"}
{"qa_id": "MOSE__7ade3068__object_world_motion_direction__02", "video_id": "MOSE__7ade3068", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (brown bull with a lighter underbelly) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/b4aa60fb14a8ce1d5428/F0.jpg", "media/b4aa60fb14a8ce1d5428/F1.jpg", "media/b4aa60fb14a8ce1d5428/F2.jpg", "media/b4aa60fb14a8ce1d5428/F3.jpg", "media/b4aa60fb14a8ce1d5428/F4.jpg", "media/b4aa60fb14a8ce1d5428/F5.jpg", "media/b4aa60fb14a8ce1d5428/F6.jpg", "media/b4aa60fb14a8ce1d5428/F7.jpg", "media/b4aa60fb14a8ce1d5428/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (brown bull with a lighter underbelly) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d87872ae250bc2a62f93e33f554cd43d95878a4a4bec926d5c0bc5cea328f873"}
{"qa_id": "MOSE__7ade3068__camera_object_distance_change__03", "video_id": "MOSE__7ade3068", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object B (brown bull with a lighter underbelly) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b4aa60fb14a8ce1d5428/F0.jpg", "media/b4aa60fb14a8ce1d5428/F1.jpg", "media/b4aa60fb14a8ce1d5428/F2.jpg", "media/b4aa60fb14a8ce1d5428/F3.jpg", "media/b4aa60fb14a8ce1d5428/F4.jpg", "media/b4aa60fb14a8ce1d5428/F5.jpg", "media/b4aa60fb14a8ce1d5428/F6.jpg", "media/b4aa60fb14a8ce1d5428/F7.jpg", "media/b4aa60fb14a8ce1d5428/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (brown bull with a lighter underbelly) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13327b0473019463aa5ed9e4458b94d8f7e8b42d5f08dd12abc7b71d1b670915"}
{"qa_id": "MOSE__7bc95dc2__camera_translation_direction__00", "video_id": "MOSE__7bc95dc2", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "down", "forward", "up"], "frame_paths": ["media/246d4255e6ad01d7ed4a/F0.jpg", "media/246d4255e6ad01d7ed4a/F1.jpg", "media/246d4255e6ad01d7ed4a/F2.jpg", "media/246d4255e6ad01d7ed4a/F3.jpg", "media/246d4255e6ad01d7ed4a/F4.jpg", "media/246d4255e6ad01d7ed4a/F5.jpg", "media/246d4255e6ad01d7ed4a/F6.jpg", "media/246d4255e6ad01d7ed4a/F7.jpg", "media/246d4255e6ad01d7ed4a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. down\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5713890b2382a643b2e4c3fcf36520bbf3d4b9fe14715b2f469b2b0ac941c020"}
{"qa_id": "MOSE__7c78e351__camera_object_distance_change__00", "video_id": "MOSE__7c78e351", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object B (white sheep with head down) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a3767eb6b2b0f2883ac3/F0.jpg", "media/a3767eb6b2b0f2883ac3/F1.jpg", "media/a3767eb6b2b0f2883ac3/F2.jpg", "media/a3767eb6b2b0f2883ac3/F3.jpg", "media/a3767eb6b2b0f2883ac3/F4.jpg", "media/a3767eb6b2b0f2883ac3/F5.jpg", "media/a3767eb6b2b0f2883ac3/F6.jpg", "media/a3767eb6b2b0f2883ac3/F7.jpg", "media/a3767eb6b2b0f2883ac3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object B (white sheep with head down) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f67793df99d7276b66421a3aeefad7dde1206e11701002759b1747a41a84d454"}
{"qa_id": "MOSE__7c78e351__object_pair_distance_change__01", "video_id": "MOSE__7c78e351", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did the 3D distance between object A (man in dark jacket crouching) and object B (white sheep with head down) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/864d8df1abd1dc8c3fbc/F0.jpg", "media/864d8df1abd1dc8c3fbc/F1.jpg", "media/864d8df1abd1dc8c3fbc/F2.jpg", "media/864d8df1abd1dc8c3fbc/F3.jpg", "media/864d8df1abd1dc8c3fbc/F4.jpg", "media/864d8df1abd1dc8c3fbc/F5.jpg", "media/864d8df1abd1dc8c3fbc/F6.jpg", "media/864d8df1abd1dc8c3fbc/F7.jpg", "media/864d8df1abd1dc8c3fbc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did the 3D distance between object A (man in dark jacket crouching) and object B (white sheep with head down) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96be05aa35c6e88a23d057ba684e0b2a0441f2a56f332d9e0513bdfa8587d9d6"}
{"qa_id": "MOSE__7d452630__object_world_motion_direction__00", "video_id": "MOSE__7d452630", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (long traditional boat with a dark roof moving on the water) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "up", "forward", "left"], "frame_paths": ["media/dd7913dd92cc2b446dc2/F0.jpg", "media/dd7913dd92cc2b446dc2/F1.jpg", "media/dd7913dd92cc2b446dc2/F2.jpg", "media/dd7913dd92cc2b446dc2/F3.jpg", "media/dd7913dd92cc2b446dc2/F4.jpg", "media/dd7913dd92cc2b446dc2/F5.jpg", "media/dd7913dd92cc2b446dc2/F6.jpg", "media/dd7913dd92cc2b446dc2/F7.jpg", "media/dd7913dd92cc2b446dc2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (long traditional boat with a dark roof moving on the water) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1495a5e2ffc990d8f94bba96266dbd6bb833d761c6457b9791298d6108a137ea"}
{"qa_id": "MOSE__7edd8b4f__camera_object_distance_change__00", "video_id": "MOSE__7edd8b4f", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (elephant in the middle of the group) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a701d7f228bd9998dbfa/F0.jpg", "media/a701d7f228bd9998dbfa/F1.jpg", "media/a701d7f228bd9998dbfa/F2.jpg", "media/a701d7f228bd9998dbfa/F3.jpg", "media/a701d7f228bd9998dbfa/F4.jpg", "media/a701d7f228bd9998dbfa/F5.jpg", "media/a701d7f228bd9998dbfa/F6.jpg", "media/a701d7f228bd9998dbfa/F7.jpg", "media/a701d7f228bd9998dbfa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (elephant in the middle of the group) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38185c9c89fe393a63d0888a753ad5f31decfa28a4c512df8e7ed9cc3e64ebdc"}
{"qa_id": "MOSE__7edd8b4f__object_pair_distance_change__01", "video_id": "MOSE__7edd8b4f", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did the 3D distance between object A (elephant on the left side of the group) and object B (elephant in the middle of the group) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/945e9136ed3339712dc7/F0.jpg", "media/945e9136ed3339712dc7/F1.jpg", "media/945e9136ed3339712dc7/F2.jpg", "media/945e9136ed3339712dc7/F3.jpg", "media/945e9136ed3339712dc7/F4.jpg", "media/945e9136ed3339712dc7/F5.jpg", "media/945e9136ed3339712dc7/F6.jpg", "media/945e9136ed3339712dc7/F7.jpg", "media/945e9136ed3339712dc7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did the 3D distance between object A (elephant on the left side of the group) and object B (elephant in the middle of the group) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af753c3a2cadea4a46a6722419f26d39613f4982fd93818062516a116a9194e1"}
{"qa_id": "MOSE__7edd8b4f__object_travel_distance_comparison__02", "video_id": "MOSE__7edd8b4f", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (elephant on the left side of the group) or B (elephant in the middle of the group)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/945e9136ed3339712dc7/F0.jpg", "media/945e9136ed3339712dc7/F1.jpg", "media/945e9136ed3339712dc7/F2.jpg", "media/945e9136ed3339712dc7/F3.jpg", "media/945e9136ed3339712dc7/F4.jpg", "media/945e9136ed3339712dc7/F5.jpg", "media/945e9136ed3339712dc7/F6.jpg", "media/945e9136ed3339712dc7/F7.jpg", "media/945e9136ed3339712dc7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (elephant on the left side of the group) or B (elephant in the middle of the group)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0fe3613f8607658d2d1b906af3f378b3c737e5323abdc2cbd7f59009e49439e4"}
{"qa_id": "MOSE__80d978e9__object_world_motion_direction__00", "video_id": "MOSE__80d978e9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "up", "backward", "left", "right"], "frame_paths": ["media/e48ac6603d98970b9429/F0.jpg", "media/e48ac6603d98970b9429/F1.jpg", "media/e48ac6603d98970b9429/F2.jpg", "media/e48ac6603d98970b9429/F3.jpg", "media/e48ac6603d98970b9429/F4.jpg", "media/e48ac6603d98970b9429/F5.jpg", "media/e48ac6603d98970b9429/F6.jpg", "media/e48ac6603d98970b9429/F7.jpg", "media/e48ac6603d98970b9429/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (orange truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65d2ddbb7d2553411033f310caf76f7cf190db4c553032b427cab51867536326"}
{"qa_id": "MOSE__80d978e9__camera_object_distance_change__01", "video_id": "MOSE__80d978e9", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (orange truck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/e48ac6603d98970b9429/F0.jpg", "media/e48ac6603d98970b9429/F1.jpg", "media/e48ac6603d98970b9429/F2.jpg", "media/e48ac6603d98970b9429/F3.jpg", "media/e48ac6603d98970b9429/F4.jpg", "media/e48ac6603d98970b9429/F5.jpg", "media/e48ac6603d98970b9429/F6.jpg", "media/e48ac6603d98970b9429/F7.jpg", "media/e48ac6603d98970b9429/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (orange truck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "243b5a01d42b66bc9e9b77db088d4aee4900f424175f33a3d70b86c98b16a9d0"}
{"qa_id": "MOSE__80d978e9__object_world_motion_direction__02", "video_id": "MOSE__80d978e9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (yellow truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "down", "backward"], "frame_paths": ["media/450f7cc2a6c8761bbabc/F0.jpg", "media/450f7cc2a6c8761bbabc/F1.jpg", "media/450f7cc2a6c8761bbabc/F2.jpg", "media/450f7cc2a6c8761bbabc/F3.jpg", "media/450f7cc2a6c8761bbabc/F4.jpg", "media/450f7cc2a6c8761bbabc/F5.jpg", "media/450f7cc2a6c8761bbabc/F6.jpg", "media/450f7cc2a6c8761bbabc/F7.jpg", "media/450f7cc2a6c8761bbabc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (yellow truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c63576aea3368a2d1861514556576f37d69cdd9ec4a5258517d2aa0805e6269"}
{"qa_id": "MOSE__80d978e9__camera_object_distance_change__03", "video_id": "MOSE__80d978e9", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (yellow truck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/450f7cc2a6c8761bbabc/F0.jpg", "media/450f7cc2a6c8761bbabc/F1.jpg", "media/450f7cc2a6c8761bbabc/F2.jpg", "media/450f7cc2a6c8761bbabc/F3.jpg", "media/450f7cc2a6c8761bbabc/F4.jpg", "media/450f7cc2a6c8761bbabc/F5.jpg", "media/450f7cc2a6c8761bbabc/F6.jpg", "media/450f7cc2a6c8761bbabc/F7.jpg", "media/450f7cc2a6c8761bbabc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (yellow truck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbb5a3c1d7096ec05e6b8426cc401d31032a6df7748844c0b35b48fb45adf05a"}
{"qa_id": "MOSE__80fb79c6__camera_object_distance_change__00", "video_id": "MOSE__80fb79c6", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (large tiger with orange and black stripes standing in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/786520631b1a931c8277/F0.jpg", "media/786520631b1a931c8277/F1.jpg", "media/786520631b1a931c8277/F2.jpg", "media/786520631b1a931c8277/F3.jpg", "media/786520631b1a931c8277/F4.jpg", "media/786520631b1a931c8277/F5.jpg", "media/786520631b1a931c8277/F6.jpg", "media/786520631b1a931c8277/F7.jpg", "media/786520631b1a931c8277/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (large tiger with orange and black stripes standing in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d3b7597e9bf12be8da6d9c16ad37f81db745318eb042382131bcf2de3a49fdd3"}
{"qa_id": "MOSE__857ddc6e__camera_translation_direction__00", "video_id": "MOSE__857ddc6e", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "left", "up", "right"], "frame_paths": ["media/5c8d1ed6f0cb126fe924/F0.jpg", "media/5c8d1ed6f0cb126fe924/F1.jpg", "media/5c8d1ed6f0cb126fe924/F2.jpg", "media/5c8d1ed6f0cb126fe924/F3.jpg", "media/5c8d1ed6f0cb126fe924/F4.jpg", "media/5c8d1ed6f0cb126fe924/F5.jpg", "media/5c8d1ed6f0cb126fe924/F6.jpg", "media/5c8d1ed6f0cb126fe924/F7.jpg", "media/5c8d1ed6f0cb126fe924/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c71a939b4aacd710e664776cd351420c00812d22195602d263ac7debc16dc282"}
{"qa_id": "MOSE__857ddc6e__object_world_motion_direction__01", "video_id": "MOSE__857ddc6e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (zebra standing near a water trough) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/b693cb7156b9a6dabe05/F0.jpg", "media/b693cb7156b9a6dabe05/F1.jpg", "media/b693cb7156b9a6dabe05/F2.jpg", "media/b693cb7156b9a6dabe05/F3.jpg", "media/b693cb7156b9a6dabe05/F4.jpg", "media/b693cb7156b9a6dabe05/F5.jpg", "media/b693cb7156b9a6dabe05/F6.jpg", "media/b693cb7156b9a6dabe05/F7.jpg", "media/b693cb7156b9a6dabe05/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (zebra standing near a water trough) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "27063fb07a20625aed0a07d119c1f56d64afc045a7e4baeb023fd9989b5f258a"}
{"qa_id": "MOSE__857ddc6e__object_pair_distance_change__02", "video_id": "MOSE__857ddc6e", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (zebra standing near a water trough) and object B (zebra standing near a tree) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/98930277f5c3ca8333f1/F0.jpg", "media/98930277f5c3ca8333f1/F1.jpg", "media/98930277f5c3ca8333f1/F2.jpg", "media/98930277f5c3ca8333f1/F3.jpg", "media/98930277f5c3ca8333f1/F4.jpg", "media/98930277f5c3ca8333f1/F5.jpg", "media/98930277f5c3ca8333f1/F6.jpg", "media/98930277f5c3ca8333f1/F7.jpg", "media/98930277f5c3ca8333f1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": ["g3", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (zebra standing near a water trough) and object B (zebra standing near a tree) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91b648f36456618767c8d691f94d3311cc25b6c823a5ee2a3fd35ed5a96dee85"}
{"qa_id": "MOSE__857e375e__camera_view_direction_change__00", "video_id": "MOSE__857e375e", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/7bd102724c200b27edb9/F0.jpg", "media/7bd102724c200b27edb9/F1.jpg", "media/7bd102724c200b27edb9/F2.jpg", "media/7bd102724c200b27edb9/F3.jpg", "media/7bd102724c200b27edb9/F4.jpg", "media/7bd102724c200b27edb9/F5.jpg", "media/7bd102724c200b27edb9/F6.jpg", "media/7bd102724c200b27edb9/F7.jpg", "media/7bd102724c200b27edb9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "MOSE__857e375e__camera_object_distance_change__01", "video_id": "MOSE__857e375e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (person in orange life vest and yellow helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ce27e9426f51bc751051/F0.jpg", "media/ce27e9426f51bc751051/F1.jpg", "media/ce27e9426f51bc751051/F2.jpg", "media/ce27e9426f51bc751051/F3.jpg", "media/ce27e9426f51bc751051/F4.jpg", "media/ce27e9426f51bc751051/F5.jpg", "media/ce27e9426f51bc751051/F6.jpg", "media/ce27e9426f51bc751051/F7.jpg", "media/ce27e9426f51bc751051/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (person in orange life vest and yellow helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a5bdb6804cce1c529252e4bb90c12534fde26938b239c51ce7a91caf040d3b6"}
{"qa_id": "MOSE__857e375e__object_world_motion_direction__02", "video_id": "MOSE__857e375e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (person in orange life vest and blue helmet) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "backward", "up", "left"], "frame_paths": ["media/616956e6444921da44a5/F0.jpg", "media/616956e6444921da44a5/F1.jpg", "media/616956e6444921da44a5/F2.jpg", "media/616956e6444921da44a5/F3.jpg", "media/616956e6444921da44a5/F4.jpg", "media/616956e6444921da44a5/F5.jpg", "media/616956e6444921da44a5/F6.jpg", "media/616956e6444921da44a5/F7.jpg", "media/616956e6444921da44a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (person in orange life vest and blue helmet) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c56618e6f7c82f986ac9e220e6e9c5fd485ea3c326b892cfee586e51f288816b"}
{"qa_id": "MOSE__85d694d7__object_world_motion_direction__00", "video_id": "MOSE__85d694d7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (man in dark suit) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "backward", "up", "left"], "frame_paths": ["media/d68e1998a8d1c76e8fc0/F0.jpg", "media/d68e1998a8d1c76e8fc0/F1.jpg", "media/d68e1998a8d1c76e8fc0/F2.jpg", "media/d68e1998a8d1c76e8fc0/F3.jpg", "media/d68e1998a8d1c76e8fc0/F4.jpg", "media/d68e1998a8d1c76e8fc0/F5.jpg", "media/d68e1998a8d1c76e8fc0/F6.jpg", "media/d68e1998a8d1c76e8fc0/F7.jpg", "media/d68e1998a8d1c76e8fc0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (man in dark suit) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa85344751041dba2c3286b209125c1e382492f888d806875161ca1ea08613c5"}
{"qa_id": "MOSE__85d694d7__camera_object_distance_change__01", "video_id": "MOSE__85d694d7", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object B (man in dark suit) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d68e1998a8d1c76e8fc0/F0.jpg", "media/d68e1998a8d1c76e8fc0/F1.jpg", "media/d68e1998a8d1c76e8fc0/F2.jpg", "media/d68e1998a8d1c76e8fc0/F3.jpg", "media/d68e1998a8d1c76e8fc0/F4.jpg", "media/d68e1998a8d1c76e8fc0/F5.jpg", "media/d68e1998a8d1c76e8fc0/F6.jpg", "media/d68e1998a8d1c76e8fc0/F7.jpg", "media/d68e1998a8d1c76e8fc0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (man in dark suit) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "001babd037995d8829fd277d1a71c77551177c8dd3c57163d7014fdbe2964581"}
{"qa_id": "MOSE__85d694d7__object_travel_distance_comparison__02", "video_id": "MOSE__85d694d7", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in orange shirt) or B (man in dark suit)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/3bd9f7608abc4da4d0d8/F0.jpg", "media/3bd9f7608abc4da4d0d8/F1.jpg", "media/3bd9f7608abc4da4d0d8/F2.jpg", "media/3bd9f7608abc4da4d0d8/F3.jpg", "media/3bd9f7608abc4da4d0d8/F4.jpg", "media/3bd9f7608abc4da4d0d8/F5.jpg", "media/3bd9f7608abc4da4d0d8/F6.jpg", "media/3bd9f7608abc4da4d0d8/F7.jpg", "media/3bd9f7608abc4da4d0d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in orange shirt) or B (man in dark suit)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cd32fec0a79721523a67645f16267dc6997525f2b26694fa7eac20c502437c03"}
{"qa_id": "MOSE__8620c024__camera_translation_direction__00", "video_id": "MOSE__8620c024", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "right", "left", "down"], "frame_paths": ["media/b277ef157594b9e9013f/F0.jpg", "media/b277ef157594b9e9013f/F1.jpg", "media/b277ef157594b9e9013f/F2.jpg", "media/b277ef157594b9e9013f/F3.jpg", "media/b277ef157594b9e9013f/F4.jpg", "media/b277ef157594b9e9013f/F5.jpg", "media/b277ef157594b9e9013f/F6.jpg", "media/b277ef157594b9e9013f/F7.jpg", "media/b277ef157594b9e9013f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bf60399d53cd05ec863be0914013db8ec13a7c31ffe848a2c39a38de36b1ba00"}
{"qa_id": "MOSE__8620c024__camera_object_distance_change__01", "video_id": "MOSE__8620c024", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (man in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fc4a9a771bba96237147/F0.jpg", "media/fc4a9a771bba96237147/F1.jpg", "media/fc4a9a771bba96237147/F2.jpg", "media/fc4a9a771bba96237147/F3.jpg", "media/fc4a9a771bba96237147/F4.jpg", "media/fc4a9a771bba96237147/F5.jpg", "media/fc4a9a771bba96237147/F6.jpg", "media/fc4a9a771bba96237147/F7.jpg", "media/fc4a9a771bba96237147/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (man in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "47e3ad96fdcbbc524e13190af8a37fce404b59f672307e46c1a7e97b242a40e0"}
{"qa_id": "MOSE__87fc565c__object_world_motion_direction__00", "video_id": "MOSE__87fc565c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (goose with light brown and white plumage) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "backward", "up"], "frame_paths": ["media/643c60a6e1212e5e4308/F0.jpg", "media/643c60a6e1212e5e4308/F1.jpg", "media/643c60a6e1212e5e4308/F2.jpg", "media/643c60a6e1212e5e4308/F3.jpg", "media/643c60a6e1212e5e4308/F4.jpg", "media/643c60a6e1212e5e4308/F5.jpg", "media/643c60a6e1212e5e4308/F6.jpg", "media/643c60a6e1212e5e4308/F7.jpg", "media/643c60a6e1212e5e4308/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (goose with light brown and white plumage) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66dc9cd786934b0cda7c031724cdb0753bb48e0cc33edaaf3768d26f8d17adfd"}
{"qa_id": "MOSE__87fc565c__object_pair_distance_change__01", "video_id": "MOSE__87fc565c", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (goose with brown and white plumage) and object B (goose with light brown and white plumage) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/edeb066b5098d8d2420c/F0.jpg", "media/edeb066b5098d8d2420c/F1.jpg", "media/edeb066b5098d8d2420c/F2.jpg", "media/edeb066b5098d8d2420c/F3.jpg", "media/edeb066b5098d8d2420c/F4.jpg", "media/edeb066b5098d8d2420c/F5.jpg", "media/edeb066b5098d8d2420c/F6.jpg", "media/edeb066b5098d8d2420c/F7.jpg", "media/edeb066b5098d8d2420c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (goose with brown and white plumage) and object B (goose with light brown and white plumage) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0891a3411edadaf506083134d1ffeb444b398dee197565e54f7923e9d6aefcf0"}
{"qa_id": "MOSE__87fc565c__object_travel_distance_comparison__02", "video_id": "MOSE__87fc565c", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (goose with brown and white plumage) or B (goose with light brown and white plumage)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/edeb066b5098d8d2420c/F0.jpg", "media/edeb066b5098d8d2420c/F1.jpg", "media/edeb066b5098d8d2420c/F2.jpg", "media/edeb066b5098d8d2420c/F3.jpg", "media/edeb066b5098d8d2420c/F4.jpg", "media/edeb066b5098d8d2420c/F5.jpg", "media/edeb066b5098d8d2420c/F6.jpg", "media/edeb066b5098d8d2420c/F7.jpg", "media/edeb066b5098d8d2420c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (goose with brown and white plumage) or B (goose with light brown and white plumage)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "00c14d5d31961e47abf43239eeb6f20be150ef922c6ee2fe7427a928108cf956"}
{"qa_id": "MOSE__89b671bb__object_world_motion_direction__00", "video_id": "MOSE__89b671bb", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (silver SUV) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/9c8dfc886e9b8a896ccf/F0.jpg", "media/9c8dfc886e9b8a896ccf/F1.jpg", "media/9c8dfc886e9b8a896ccf/F2.jpg", "media/9c8dfc886e9b8a896ccf/F3.jpg", "media/9c8dfc886e9b8a896ccf/F4.jpg", "media/9c8dfc886e9b8a896ccf/F5.jpg", "media/9c8dfc886e9b8a896ccf/F6.jpg", "media/9c8dfc886e9b8a896ccf/F7.jpg", "media/9c8dfc886e9b8a896ccf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (silver SUV) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78283b8fad95b3707e65198d99f86d858d0f5effd96fd08970e0f5f5349d7ed4"}
{"qa_id": "MOSE__89b671bb__object_pair_distance_change__01", "video_id": "MOSE__89b671bb", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did the 3D distance between object A (silver SUV) and object B (person riding a scooter) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/bba7c1cfc38e8b20e8ef/F0.jpg", "media/bba7c1cfc38e8b20e8ef/F1.jpg", "media/bba7c1cfc38e8b20e8ef/F2.jpg", "media/bba7c1cfc38e8b20e8ef/F3.jpg", "media/bba7c1cfc38e8b20e8ef/F4.jpg", "media/bba7c1cfc38e8b20e8ef/F5.jpg", "media/bba7c1cfc38e8b20e8ef/F6.jpg", "media/bba7c1cfc38e8b20e8ef/F7.jpg", "media/bba7c1cfc38e8b20e8ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (silver SUV) and object B (person riding a scooter) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c95876753d4ea84776be306ecfe35e5c3bcae336463f25934f84ab793764dc1d"}
{"qa_id": "MOSE__89b671bb__object_travel_distance_comparison__02", "video_id": "MOSE__89b671bb", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (silver SUV) or B (person riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/bba7c1cfc38e8b20e8ef/F0.jpg", "media/bba7c1cfc38e8b20e8ef/F1.jpg", "media/bba7c1cfc38e8b20e8ef/F2.jpg", "media/bba7c1cfc38e8b20e8ef/F3.jpg", "media/bba7c1cfc38e8b20e8ef/F4.jpg", "media/bba7c1cfc38e8b20e8ef/F5.jpg", "media/bba7c1cfc38e8b20e8ef/F6.jpg", "media/bba7c1cfc38e8b20e8ef/F7.jpg", "media/bba7c1cfc38e8b20e8ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (silver SUV) or B (person riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b6caa6a8e3fa13ac6871065f67922196c0f4a3d6eced601f710846aee1d35151"}
{"qa_id": "MOSE__8c494902__camera_view_direction_change__00", "video_id": "MOSE__8c494902", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/c078d91fbfe63acf20a3/F0.jpg", "media/c078d91fbfe63acf20a3/F1.jpg", "media/c078d91fbfe63acf20a3/F2.jpg", "media/c078d91fbfe63acf20a3/F3.jpg", "media/c078d91fbfe63acf20a3/F4.jpg", "media/c078d91fbfe63acf20a3/F5.jpg", "media/c078d91fbfe63acf20a3/F6.jpg", "media/c078d91fbfe63acf20a3/F7.jpg", "media/c078d91fbfe63acf20a3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3088edbf9f3f4f0a21f71877cf93ce4c2687ff31dc9648f2eed7eb10be96f6f6"}
{"qa_id": "MOSE__8c8b7913__object_pair_distance_change__00", "video_id": "MOSE__8c8b7913", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F7, did the 3D distance between object A (light brown calf with a white face) and object B (white calf with a dark face) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/71bec1fce71c0b800306/F0.jpg", "media/71bec1fce71c0b800306/F1.jpg", "media/71bec1fce71c0b800306/F2.jpg", "media/71bec1fce71c0b800306/F3.jpg", "media/71bec1fce71c0b800306/F4.jpg", "media/71bec1fce71c0b800306/F5.jpg", "media/71bec1fce71c0b800306/F6.jpg", "media/71bec1fce71c0b800306/F7.jpg", "media/71bec1fce71c0b800306/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did the 3D distance between object A (light brown calf with a white face) and object B (white calf with a dark face) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2a39d49b8b12a0b06306fa22b76f6655ed6994b2e1b1f626370e20377ecb402"}
{"qa_id": "MOSE__8c8b7913__object_travel_distance_comparison__01", "video_id": "MOSE__8c8b7913", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown calf with a white face) or B (white calf with a dark face)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/71bec1fce71c0b800306/F0.jpg", "media/71bec1fce71c0b800306/F1.jpg", "media/71bec1fce71c0b800306/F2.jpg", "media/71bec1fce71c0b800306/F3.jpg", "media/71bec1fce71c0b800306/F4.jpg", "media/71bec1fce71c0b800306/F5.jpg", "media/71bec1fce71c0b800306/F6.jpg", "media/71bec1fce71c0b800306/F7.jpg", "media/71bec1fce71c0b800306/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown calf with a white face) or B (white calf with a dark face)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "307c700f56b69dcca4a3132011290f35a50c2a4b69a0459f08b391fcf983ba42"}
{"qa_id": "MOSE__8cf49218__object_world_motion_direction__00", "video_id": "MOSE__8cf49218", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (zebra walking to the right) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "left", "down", "right"], "frame_paths": ["media/6ebebc5deaf4e324390a/F0.jpg", "media/6ebebc5deaf4e324390a/F1.jpg", "media/6ebebc5deaf4e324390a/F2.jpg", "media/6ebebc5deaf4e324390a/F3.jpg", "media/6ebebc5deaf4e324390a/F4.jpg", "media/6ebebc5deaf4e324390a/F5.jpg", "media/6ebebc5deaf4e324390a/F6.jpg", "media/6ebebc5deaf4e324390a/F7.jpg", "media/6ebebc5deaf4e324390a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (zebra walking to the right) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2261abdd220e1a34c11d5c709f714ff79ba6f78a3ba1f3e2dea996b717d32f1c"}
{"qa_id": "MOSE__8d23ddcc__object_pair_distance_change__00", "video_id": "MOSE__8d23ddcc", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (orange truck with a boxy cargo area) and object B (dark green truck with a cab and cargo bed) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/cfd3e02de1c1430cf4f6/F0.jpg", "media/cfd3e02de1c1430cf4f6/F1.jpg", "media/cfd3e02de1c1430cf4f6/F2.jpg", "media/cfd3e02de1c1430cf4f6/F3.jpg", "media/cfd3e02de1c1430cf4f6/F4.jpg", "media/cfd3e02de1c1430cf4f6/F5.jpg", "media/cfd3e02de1c1430cf4f6/F6.jpg", "media/cfd3e02de1c1430cf4f6/F7.jpg", "media/cfd3e02de1c1430cf4f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (orange truck with a boxy cargo area) and object B (dark green truck with a cab and cargo bed) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "984961f8c8093ed61b5e8768f369671b6c86815de1a16b61bb60b6d10c0a78bb"}
{"qa_id": "MOSE__8d2ddeff__object_world_motion_direction__00", "video_id": "MOSE__8d2ddeff", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (red and blue city bus) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "left", "up", "forward"], "frame_paths": ["media/fba3afac3f3e724ebb68/F0.jpg", "media/fba3afac3f3e724ebb68/F1.jpg", "media/fba3afac3f3e724ebb68/F2.jpg", "media/fba3afac3f3e724ebb68/F3.jpg", "media/fba3afac3f3e724ebb68/F4.jpg", "media/fba3afac3f3e724ebb68/F5.jpg", "media/fba3afac3f3e724ebb68/F6.jpg", "media/fba3afac3f3e724ebb68/F7.jpg", "media/fba3afac3f3e724ebb68/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (red and blue city bus) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dab6c93533e69df696735f5d21274a29d14b3e1971f8e09e03a3136884c04054"}
{"qa_id": "MOSE__8d2ddeff__object_pair_distance_change__01", "video_id": "MOSE__8d2ddeff", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F8, did the 3D distance between object A (red and blue city bus) and object B (person in light blue shirt riding a scooter) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/549d9d660b5cf00b29d4/F0.jpg", "media/549d9d660b5cf00b29d4/F1.jpg", "media/549d9d660b5cf00b29d4/F2.jpg", "media/549d9d660b5cf00b29d4/F3.jpg", "media/549d9d660b5cf00b29d4/F4.jpg", "media/549d9d660b5cf00b29d4/F5.jpg", "media/549d9d660b5cf00b29d4/F6.jpg", "media/549d9d660b5cf00b29d4/F7.jpg", "media/549d9d660b5cf00b29d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (red and blue city bus) and object B (person in light blue shirt riding a scooter) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cf6a71135e8dbb046412e3a560c9ad5cc58880bd7fa3fcba63a94ad1fa3647c9"}
{"qa_id": "MOSE__8d2ddeff__object_travel_distance_comparison__02", "video_id": "MOSE__8d2ddeff", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (red and blue city bus) or B (person in light blue shirt riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/549d9d660b5cf00b29d4/F0.jpg", "media/549d9d660b5cf00b29d4/F1.jpg", "media/549d9d660b5cf00b29d4/F2.jpg", "media/549d9d660b5cf00b29d4/F3.jpg", "media/549d9d660b5cf00b29d4/F4.jpg", "media/549d9d660b5cf00b29d4/F5.jpg", "media/549d9d660b5cf00b29d4/F6.jpg", "media/549d9d660b5cf00b29d4/F7.jpg", "media/549d9d660b5cf00b29d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (red and blue city bus) or B (person in light blue shirt riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dab10f8fd776c744d8ca3554695050f296c602feff3a2a10087ff26af090f93a"}
{"qa_id": "MOSE__8dbebbd9__camera_translation_direction__00", "video_id": "MOSE__8dbebbd9", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "backward", "left", "down", "forward"], "frame_paths": ["media/26c16c0f039cb1406b74/F0.jpg", "media/26c16c0f039cb1406b74/F1.jpg", "media/26c16c0f039cb1406b74/F2.jpg", "media/26c16c0f039cb1406b74/F3.jpg", "media/26c16c0f039cb1406b74/F4.jpg", "media/26c16c0f039cb1406b74/F5.jpg", "media/26c16c0f039cb1406b74/F6.jpg", "media/26c16c0f039cb1406b74/F7.jpg", "media/26c16c0f039cb1406b74/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. backward\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8149f873b4f09d437059a518e70c5c43aa0dcc069da509b2bbeff7e2e89b2024"}
{"qa_id": "MOSE__8fa50320__camera_translation_direction__00", "video_id": "MOSE__8fa50320", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/ec76c74c3c5cfaa0756a/F0.jpg", "media/ec76c74c3c5cfaa0756a/F1.jpg", "media/ec76c74c3c5cfaa0756a/F2.jpg", "media/ec76c74c3c5cfaa0756a/F3.jpg", "media/ec76c74c3c5cfaa0756a/F4.jpg", "media/ec76c74c3c5cfaa0756a/F5.jpg", "media/ec76c74c3c5cfaa0756a/F6.jpg", "media/ec76c74c3c5cfaa0756a/F7.jpg", "media/ec76c74c3c5cfaa0756a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2c450acf2fb1beb0e884ec389f961b1606edafa5dee49c36ccda0700a910684e"}
{"qa_id": "MOSE__8fa50320__camera_view_direction_change__01", "video_id": "MOSE__8fa50320", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/ec76c74c3c5cfaa0756a/F0.jpg", "media/ec76c74c3c5cfaa0756a/F1.jpg", "media/ec76c74c3c5cfaa0756a/F2.jpg", "media/ec76c74c3c5cfaa0756a/F3.jpg", "media/ec76c74c3c5cfaa0756a/F4.jpg", "media/ec76c74c3c5cfaa0756a/F5.jpg", "media/ec76c74c3c5cfaa0756a/F6.jpg", "media/ec76c74c3c5cfaa0756a/F7.jpg", "media/ec76c74c3c5cfaa0756a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5d47aef5ded49513cb794135dfce74960c5515464cb16527eec00058b2c9cfe"}
{"qa_id": "MOSE__8fa50320__camera_object_distance_change__02", "video_id": "MOSE__8fa50320", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (large elephant in the foreground walking left) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/3ca0f38898867f7dec78/F0.jpg", "media/3ca0f38898867f7dec78/F1.jpg", "media/3ca0f38898867f7dec78/F2.jpg", "media/3ca0f38898867f7dec78/F3.jpg", "media/3ca0f38898867f7dec78/F4.jpg", "media/3ca0f38898867f7dec78/F5.jpg", "media/3ca0f38898867f7dec78/F6.jpg", "media/3ca0f38898867f7dec78/F7.jpg", "media/3ca0f38898867f7dec78/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (large elephant in the foreground walking left) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a8d319315039114de8c815e106de057ca79c38da17da41365123cd232935eab"}
{"qa_id": "MOSE__8fa50320__object_pair_distance_change__03", "video_id": "MOSE__8fa50320", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did the 3D distance between object A (large elephant in the foreground walking left) and object B (elephant walking to the right of the foreground elephant) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/078792558512382e75bf/F0.jpg", "media/078792558512382e75bf/F1.jpg", "media/078792558512382e75bf/F2.jpg", "media/078792558512382e75bf/F3.jpg", "media/078792558512382e75bf/F4.jpg", "media/078792558512382e75bf/F5.jpg", "media/078792558512382e75bf/F6.jpg", "media/078792558512382e75bf/F7.jpg", "media/078792558512382e75bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (large elephant in the foreground walking left) and object B (elephant walking to the right of the foreground elephant) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "83dccf828240f50e955870adc0052a89d385af5569e83ed7471d827e17b06a42"}
{"qa_id": "MOSE__9113c9d4__object_world_motion_direction__00", "video_id": "MOSE__9113c9d4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (light brown chicken with dark tail feathers) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/ce190909150cbf0e36cc/F0.jpg", "media/ce190909150cbf0e36cc/F1.jpg", "media/ce190909150cbf0e36cc/F2.jpg", "media/ce190909150cbf0e36cc/F3.jpg", "media/ce190909150cbf0e36cc/F4.jpg", "media/ce190909150cbf0e36cc/F5.jpg", "media/ce190909150cbf0e36cc/F6.jpg", "media/ce190909150cbf0e36cc/F7.jpg", "media/ce190909150cbf0e36cc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (light brown chicken with dark tail feathers) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34337677e1eb85ad5416b23515df76bcf262641e7a740ce7d558639de77874af"}
{"qa_id": "MOSE__9113c9d4__object_pair_distance_change__01", "video_id": "MOSE__9113c9d4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (light brown chicken with dark tail feathers) and object B (reddish-brown chicken pecking at a white bowl) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/951fe459ada79dab81d9/F0.jpg", "media/951fe459ada79dab81d9/F1.jpg", "media/951fe459ada79dab81d9/F2.jpg", "media/951fe459ada79dab81d9/F3.jpg", "media/951fe459ada79dab81d9/F4.jpg", "media/951fe459ada79dab81d9/F5.jpg", "media/951fe459ada79dab81d9/F6.jpg", "media/951fe459ada79dab81d9/F7.jpg", "media/951fe459ada79dab81d9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (light brown chicken with dark tail feathers) and object B (reddish-brown chicken pecking at a white bowl) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aaabb3fafe6a32d6a4b963e7f98d0319e2bd768c4c4d57f8b93a0debe5add0b1"}
{"qa_id": "MOSE__9113c9d4__object_travel_distance_comparison__02", "video_id": "MOSE__9113c9d4", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown chicken with dark tail feathers) or B (reddish-brown chicken pecking at a white bowl)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/951fe459ada79dab81d9/F0.jpg", "media/951fe459ada79dab81d9/F1.jpg", "media/951fe459ada79dab81d9/F2.jpg", "media/951fe459ada79dab81d9/F3.jpg", "media/951fe459ada79dab81d9/F4.jpg", "media/951fe459ada79dab81d9/F5.jpg", "media/951fe459ada79dab81d9/F6.jpg", "media/951fe459ada79dab81d9/F7.jpg", "media/951fe459ada79dab81d9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown chicken with dark tail feathers) or B (reddish-brown chicken pecking at a white bowl)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb6750503a6fd9f623d406150edff1f91543b5f3140f41a1e381e5ca1c29507d"}
{"qa_id": "MOSE__915f0d9a__object_world_motion_direction__00", "video_id": "MOSE__915f0d9a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "left", "right", "backward"], "frame_paths": ["media/7b8a831eb25bf058520a/F0.jpg", "media/7b8a831eb25bf058520a/F1.jpg", "media/7b8a831eb25bf058520a/F2.jpg", "media/7b8a831eb25bf058520a/F3.jpg", "media/7b8a831eb25bf058520a/F4.jpg", "media/7b8a831eb25bf058520a/F5.jpg", "media/7b8a831eb25bf058520a/F6.jpg", "media/7b8a831eb25bf058520a/F7.jpg", "media/7b8a831eb25bf058520a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8de6d0027ff92f965f015c3b57f80203f754146375b903d70e901dd2f49326ec"}
{"qa_id": "MOSE__915f0d9a__object_world_motion_direction__01", "video_id": "MOSE__915f0d9a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (smaller elephant walking behind the large adult) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/70609d2c71c7ef714ab3/F0.jpg", "media/70609d2c71c7ef714ab3/F1.jpg", "media/70609d2c71c7ef714ab3/F2.jpg", "media/70609d2c71c7ef714ab3/F3.jpg", "media/70609d2c71c7ef714ab3/F4.jpg", "media/70609d2c71c7ef714ab3/F5.jpg", "media/70609d2c71c7ef714ab3/F6.jpg", "media/70609d2c71c7ef714ab3/F7.jpg", "media/70609d2c71c7ef714ab3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (smaller elephant walking behind the large adult) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0c3c2bdd00ff8e865d008f2be9504552fc4cc0d1753900fa61cfb10da917ab94"}
{"qa_id": "MOSE__91b754c5__camera_translation_direction__00", "video_id": "MOSE__91b754c5", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "down", "up"], "frame_paths": ["media/734d7319b1dad3425034/F0.jpg", "media/734d7319b1dad3425034/F1.jpg", "media/734d7319b1dad3425034/F2.jpg", "media/734d7319b1dad3425034/F3.jpg", "media/734d7319b1dad3425034/F4.jpg", "media/734d7319b1dad3425034/F5.jpg", "media/734d7319b1dad3425034/F6.jpg", "media/734d7319b1dad3425034/F7.jpg", "media/734d7319b1dad3425034/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "037dc1eaa8b508d2b1ffec8694f46d33c3d4a7d6f70d646b7637ff3402ec4a60"}
{"qa_id": "MOSE__9293b755__object_world_motion_direction__00", "video_id": "MOSE__9293b755", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and black shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "forward", "right", "left", "down"], "frame_paths": ["media/6788352ee414052a76e8/F0.jpg", "media/6788352ee414052a76e8/F1.jpg", "media/6788352ee414052a76e8/F2.jpg", "media/6788352ee414052a76e8/F3.jpg", "media/6788352ee414052a76e8/F4.jpg", "media/6788352ee414052a76e8/F5.jpg", "media/6788352ee414052a76e8/F6.jpg", "media/6788352ee414052a76e8/F7.jpg", "media/6788352ee414052a76e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and black shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6745da3bd5db756c479040172de50b36b10f3d3c605ab27aeb1119d8b58d7f48"}
{"qa_id": "MOSE__9293b755__object_pair_distance_change__01", "video_id": "MOSE__9293b755", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (person in blue shirt and black shorts) and object B (person in red shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/fb69360007e6ae5812ee/F0.jpg", "media/fb69360007e6ae5812ee/F1.jpg", "media/fb69360007e6ae5812ee/F2.jpg", "media/fb69360007e6ae5812ee/F3.jpg", "media/fb69360007e6ae5812ee/F4.jpg", "media/fb69360007e6ae5812ee/F5.jpg", "media/fb69360007e6ae5812ee/F6.jpg", "media/fb69360007e6ae5812ee/F7.jpg", "media/fb69360007e6ae5812ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (person in blue shirt and black shorts) and object B (person in red shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1debdd80336ac572178cf49d359a3a545f4b57a38a8caff0d4a8f1e3ae5cf89"}
{"qa_id": "MOSE__9293b755__object_travel_distance_comparison__02", "video_id": "MOSE__9293b755", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and black shorts) or B (person in red shirt and black shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/fb69360007e6ae5812ee/F0.jpg", "media/fb69360007e6ae5812ee/F1.jpg", "media/fb69360007e6ae5812ee/F2.jpg", "media/fb69360007e6ae5812ee/F3.jpg", "media/fb69360007e6ae5812ee/F4.jpg", "media/fb69360007e6ae5812ee/F5.jpg", "media/fb69360007e6ae5812ee/F6.jpg", "media/fb69360007e6ae5812ee/F7.jpg", "media/fb69360007e6ae5812ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and black shorts) or B (person in red shirt and black shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "754d6cb0b2ad6705aba475caa27a619f6173c8f8ab7c2692a0566b9d06b431c9"}
{"qa_id": "MOSE__932542cd__object_pair_distance_change__00", "video_id": "MOSE__932542cd", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F5, did the 3D distance between object A (giraffe standing in the center group) and object B (giraffe standing to the right of the center group) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/132eef75c98a9013dc02/F0.jpg", "media/132eef75c98a9013dc02/F1.jpg", "media/132eef75c98a9013dc02/F2.jpg", "media/132eef75c98a9013dc02/F3.jpg", "media/132eef75c98a9013dc02/F4.jpg", "media/132eef75c98a9013dc02/F5.jpg", "media/132eef75c98a9013dc02/F6.jpg", "media/132eef75c98a9013dc02/F7.jpg", "media/132eef75c98a9013dc02/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (giraffe standing in the center group) and object B (giraffe standing to the right of the center group) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4cbdee5ca7809dd80554fcfc6b3980981ecc9add751f526d74ab8002e8813d34"}
{"qa_id": "MOSE__9421b9ad__camera_view_direction_change__00", "video_id": "MOSE__9421b9ad", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/6372a76c698dd1857ee3/F0.jpg", "media/6372a76c698dd1857ee3/F1.jpg", "media/6372a76c698dd1857ee3/F2.jpg", "media/6372a76c698dd1857ee3/F3.jpg", "media/6372a76c698dd1857ee3/F4.jpg", "media/6372a76c698dd1857ee3/F5.jpg", "media/6372a76c698dd1857ee3/F6.jpg", "media/6372a76c698dd1857ee3/F7.jpg", "media/6372a76c698dd1857ee3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88b40c1ef52643cefe4ef1c3fa249f1a470c7fcca2783a307d04d9b65d9496e2"}
{"qa_id": "MOSE__94767cb4__object_world_motion_direction__00", "video_id": "MOSE__94767cb4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (small bird with orange head and green body) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "left", "up", "forward", "right"], "frame_paths": ["media/889215ee3f3e84a6753a/F0.jpg", "media/889215ee3f3e84a6753a/F1.jpg", "media/889215ee3f3e84a6753a/F2.jpg", "media/889215ee3f3e84a6753a/F3.jpg", "media/889215ee3f3e84a6753a/F4.jpg", "media/889215ee3f3e84a6753a/F5.jpg", "media/889215ee3f3e84a6753a/F6.jpg", "media/889215ee3f3e84a6753a/F7.jpg", "media/889215ee3f3e84a6753a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (small bird with orange head and green body) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. left\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a66ec406e8d2eeac1ce31db1bc586bf3886c1cc29720173429d7ce3fe7944f32"}
{"qa_id": "MOSE__94767cb4__object_pair_distance_change__01", "video_id": "MOSE__94767cb4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (small bird with orange head and green body) and object B (small grey bird) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/9775dd7cab50cfe17924/F0.jpg", "media/9775dd7cab50cfe17924/F1.jpg", "media/9775dd7cab50cfe17924/F2.jpg", "media/9775dd7cab50cfe17924/F3.jpg", "media/9775dd7cab50cfe17924/F4.jpg", "media/9775dd7cab50cfe17924/F5.jpg", "media/9775dd7cab50cfe17924/F6.jpg", "media/9775dd7cab50cfe17924/F7.jpg", "media/9775dd7cab50cfe17924/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (small bird with orange head and green body) and object B (small grey bird) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3f1824664e9a67b8efe36672387c9b40e6250b89fde610e665b85d72ad51cd60"}
{"qa_id": "MOSE__94767cb4__object_travel_distance_comparison__02", "video_id": "MOSE__94767cb4", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small bird with orange head and green body) or B (small grey bird)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/9775dd7cab50cfe17924/F0.jpg", "media/9775dd7cab50cfe17924/F1.jpg", "media/9775dd7cab50cfe17924/F2.jpg", "media/9775dd7cab50cfe17924/F3.jpg", "media/9775dd7cab50cfe17924/F4.jpg", "media/9775dd7cab50cfe17924/F5.jpg", "media/9775dd7cab50cfe17924/F6.jpg", "media/9775dd7cab50cfe17924/F7.jpg", "media/9775dd7cab50cfe17924/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small bird with orange head and green body) or B (small grey bird)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "990f39e7fca699c3f179502f216acc6a534414dde646cbc824cf69c8570dff09"}
{"qa_id": "MOSE__94d3403a__camera_view_direction_change__00", "video_id": "MOSE__94d3403a", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/307b6adb84efd053be92/F0.jpg", "media/307b6adb84efd053be92/F1.jpg", "media/307b6adb84efd053be92/F2.jpg", "media/307b6adb84efd053be92/F3.jpg", "media/307b6adb84efd053be92/F4.jpg", "media/307b6adb84efd053be92/F5.jpg", "media/307b6adb84efd053be92/F6.jpg", "media/307b6adb84efd053be92/F7.jpg", "media/307b6adb84efd053be92/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dcea6da4b2f031518ed8c356ea0f756ba63782cf351f89f1f2d49381c78deedf"}
{"qa_id": "MOSE__94d3403a__object_world_motion_direction__01", "video_id": "MOSE__94d3403a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (white SUV) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "up", "forward", "down"], "frame_paths": ["media/30e77d7db8d470cc806f/F0.jpg", "media/30e77d7db8d470cc806f/F1.jpg", "media/30e77d7db8d470cc806f/F2.jpg", "media/30e77d7db8d470cc806f/F3.jpg", "media/30e77d7db8d470cc806f/F4.jpg", "media/30e77d7db8d470cc806f/F5.jpg", "media/30e77d7db8d470cc806f/F6.jpg", "media/30e77d7db8d470cc806f/F7.jpg", "media/30e77d7db8d470cc806f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (white SUV) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c659d334f9bd6c1fa2733b5c91d22f7789ba8d0dde03117da16d84ce9cae51f0"}
{"qa_id": "MOSE__94d3403a__camera_object_distance_change__02", "video_id": "MOSE__94d3403a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/30e77d7db8d470cc806f/F0.jpg", "media/30e77d7db8d470cc806f/F1.jpg", "media/30e77d7db8d470cc806f/F2.jpg", "media/30e77d7db8d470cc806f/F3.jpg", "media/30e77d7db8d470cc806f/F4.jpg", "media/30e77d7db8d470cc806f/F5.jpg", "media/30e77d7db8d470cc806f/F6.jpg", "media/30e77d7db8d470cc806f/F7.jpg", "media/30e77d7db8d470cc806f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (white SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a3668cd8c5dde79c427b48ab30fbb516c667cd562d95bd6413d34842a3f5a077"}
{"qa_id": "MOSE__95215aa1__object_world_motion_direction__00", "video_id": "MOSE__95215aa1", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (grey pigeon with black wing bars) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/fa9d15f6fc3cb9d20b6b/F0.jpg", "media/fa9d15f6fc3cb9d20b6b/F1.jpg", "media/fa9d15f6fc3cb9d20b6b/F2.jpg", "media/fa9d15f6fc3cb9d20b6b/F3.jpg", "media/fa9d15f6fc3cb9d20b6b/F4.jpg", "media/fa9d15f6fc3cb9d20b6b/F5.jpg", "media/fa9d15f6fc3cb9d20b6b/F6.jpg", "media/fa9d15f6fc3cb9d20b6b/F7.jpg", "media/fa9d15f6fc3cb9d20b6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (grey pigeon with black wing bars) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "30ab1b0874b180a05334c3d233fa8dd1c84cefc9b75c9c01a5781333ccfe2cc3"}
{"qa_id": "MOSE__95cf72d7__camera_translation_direction__00", "video_id": "MOSE__95cf72d7", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "backward", "up"], "frame_paths": ["media/c80d10dca9c7765e2a80/F0.jpg", "media/c80d10dca9c7765e2a80/F1.jpg", "media/c80d10dca9c7765e2a80/F2.jpg", "media/c80d10dca9c7765e2a80/F3.jpg", "media/c80d10dca9c7765e2a80/F4.jpg", "media/c80d10dca9c7765e2a80/F5.jpg", "media/c80d10dca9c7765e2a80/F6.jpg", "media/c80d10dca9c7765e2a80/F7.jpg", "media/c80d10dca9c7765e2a80/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca03bab18ff00f6537bb1b1127440e5784cf98a6e27cc74e64f75d478f59f253"}
{"qa_id": "MOSE__95e24093__object_world_motion_direction__00", "video_id": "MOSE__95e24093", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (solid brown horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "right", "down"], "frame_paths": ["media/e0e730aaef00c25b6d5a/F0.jpg", "media/e0e730aaef00c25b6d5a/F1.jpg", "media/e0e730aaef00c25b6d5a/F2.jpg", "media/e0e730aaef00c25b6d5a/F3.jpg", "media/e0e730aaef00c25b6d5a/F4.jpg", "media/e0e730aaef00c25b6d5a/F5.jpg", "media/e0e730aaef00c25b6d5a/F6.jpg", "media/e0e730aaef00c25b6d5a/F7.jpg", "media/e0e730aaef00c25b6d5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (solid brown horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e556ba6973517e4fab49e9d9681113c960daac46c4c55deebdf1338cdcbedf2a"}
{"qa_id": "MOSE__95e24093__object_world_motion_direction__01", "video_id": "MOSE__95e24093", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (distant horse in background) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "backward", "forward", "left", "up"], "frame_paths": ["media/62401afeab63a57fab60/F0.jpg", "media/62401afeab63a57fab60/F1.jpg", "media/62401afeab63a57fab60/F2.jpg", "media/62401afeab63a57fab60/F3.jpg", "media/62401afeab63a57fab60/F4.jpg", "media/62401afeab63a57fab60/F5.jpg", "media/62401afeab63a57fab60/F6.jpg", "media/62401afeab63a57fab60/F7.jpg", "media/62401afeab63a57fab60/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (distant horse in background) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. backward\nD. forward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "03bfd0d29363ae7b72f5d6214945e57eb2e28cf48d2d5f21bde5667c6fe7b202"}
{"qa_id": "MOSE__95e24093__camera_object_distance_change__02", "video_id": "MOSE__95e24093", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (distant horse in background) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/62401afeab63a57fab60/F0.jpg", "media/62401afeab63a57fab60/F1.jpg", "media/62401afeab63a57fab60/F2.jpg", "media/62401afeab63a57fab60/F3.jpg", "media/62401afeab63a57fab60/F4.jpg", "media/62401afeab63a57fab60/F5.jpg", "media/62401afeab63a57fab60/F6.jpg", "media/62401afeab63a57fab60/F7.jpg", "media/62401afeab63a57fab60/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (distant horse in background) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f0409cf43ab2011504d9341829b5d6cf5e4ef4e140f0bb3975e6696ed8013f15"}
{"qa_id": "MOSE__95e24093__object_pair_distance_change__03", "video_id": "MOSE__95e24093", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (solid brown horse) and object B (distant horse in background) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/f52c68c1cbb96b03f04a/F0.jpg", "media/f52c68c1cbb96b03f04a/F1.jpg", "media/f52c68c1cbb96b03f04a/F2.jpg", "media/f52c68c1cbb96b03f04a/F3.jpg", "media/f52c68c1cbb96b03f04a/F4.jpg", "media/f52c68c1cbb96b03f04a/F5.jpg", "media/f52c68c1cbb96b03f04a/F6.jpg", "media/f52c68c1cbb96b03f04a/F7.jpg", "media/f52c68c1cbb96b03f04a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (solid brown horse) and object B (distant horse in background) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f5377e59443dad93231230e789efd5e101ebc0047ae03cc06f689c5f938b143"}
{"qa_id": "MOSE__95e24093__object_travel_distance_comparison__04", "video_id": "MOSE__95e24093", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (solid brown horse) or B (distant horse in background)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/f52c68c1cbb96b03f04a/F0.jpg", "media/f52c68c1cbb96b03f04a/F1.jpg", "media/f52c68c1cbb96b03f04a/F2.jpg", "media/f52c68c1cbb96b03f04a/F3.jpg", "media/f52c68c1cbb96b03f04a/F4.jpg", "media/f52c68c1cbb96b03f04a/F5.jpg", "media/f52c68c1cbb96b03f04a/F6.jpg", "media/f52c68c1cbb96b03f04a/F7.jpg", "media/f52c68c1cbb96b03f04a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (solid brown horse) or B (distant horse in background)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b25f408ba1b88f379ce59ffef9a426be9ff4889cd28ccd473fe24d5cbcc962e2"}
{"qa_id": "MOSE__97b0811e__camera_view_direction_change__00", "video_id": "MOSE__97b0811e", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/68baf8c3e9395ca28c45/F0.jpg", "media/68baf8c3e9395ca28c45/F1.jpg", "media/68baf8c3e9395ca28c45/F2.jpg", "media/68baf8c3e9395ca28c45/F3.jpg", "media/68baf8c3e9395ca28c45/F4.jpg", "media/68baf8c3e9395ca28c45/F5.jpg", "media/68baf8c3e9395ca28c45/F6.jpg", "media/68baf8c3e9395ca28c45/F7.jpg", "media/68baf8c3e9395ca28c45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "MOSE__97b0811e__object_world_motion_direction__01", "video_id": "MOSE__97b0811e", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (large adult elephant walking to the right) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/ab8d53c25621268b85fa/F0.jpg", "media/ab8d53c25621268b85fa/F1.jpg", "media/ab8d53c25621268b85fa/F2.jpg", "media/ab8d53c25621268b85fa/F3.jpg", "media/ab8d53c25621268b85fa/F4.jpg", "media/ab8d53c25621268b85fa/F5.jpg", "media/ab8d53c25621268b85fa/F6.jpg", "media/ab8d53c25621268b85fa/F7.jpg", "media/ab8d53c25621268b85fa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (large adult elephant walking to the right) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe540dc17b068768a71c0237e93222cfd2e33455f90270dda84daee3c52bcf43"}
{"qa_id": "MOSE__97b0811e__camera_object_distance_change__02", "video_id": "MOSE__97b0811e", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did object A (large adult elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ab8d53c25621268b85fa/F0.jpg", "media/ab8d53c25621268b85fa/F1.jpg", "media/ab8d53c25621268b85fa/F2.jpg", "media/ab8d53c25621268b85fa/F3.jpg", "media/ab8d53c25621268b85fa/F4.jpg", "media/ab8d53c25621268b85fa/F5.jpg", "media/ab8d53c25621268b85fa/F6.jpg", "media/ab8d53c25621268b85fa/F7.jpg", "media/ab8d53c25621268b85fa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (large adult elephant walking to the right) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7acd110f44e85f030cbfd5034186dee8939207fe7efe75813f05eae2fda048e4"}
{"qa_id": "MOSE__97b0811e__object_world_motion_direction__03", "video_id": "MOSE__97b0811e", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (small juvenile elephant walking to the left) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "down", "up"], "frame_paths": ["media/d909b88be713750bcf7d/F0.jpg", "media/d909b88be713750bcf7d/F1.jpg", "media/d909b88be713750bcf7d/F2.jpg", "media/d909b88be713750bcf7d/F3.jpg", "media/d909b88be713750bcf7d/F4.jpg", "media/d909b88be713750bcf7d/F5.jpg", "media/d909b88be713750bcf7d/F6.jpg", "media/d909b88be713750bcf7d/F7.jpg", "media/d909b88be713750bcf7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (small juvenile elephant walking to the left) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1d9c072bf75ec6d3c47e851f6f6159404017176d0b5066125f2bf5a7ebadb5a2"}
{"qa_id": "MOSE__97b0811e__camera_object_distance_change__04", "video_id": "MOSE__97b0811e", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did object B (small juvenile elephant walking to the left) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d909b88be713750bcf7d/F0.jpg", "media/d909b88be713750bcf7d/F1.jpg", "media/d909b88be713750bcf7d/F2.jpg", "media/d909b88be713750bcf7d/F3.jpg", "media/d909b88be713750bcf7d/F4.jpg", "media/d909b88be713750bcf7d/F5.jpg", "media/d909b88be713750bcf7d/F6.jpg", "media/d909b88be713750bcf7d/F7.jpg", "media/d909b88be713750bcf7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (small juvenile elephant walking to the left) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "60751dd1863abcfafb32d1d4dd549621e02633a618fe657603e06066c537e597"}
{"qa_id": "MOSE__97b0811e__object_pair_distance_change__05", "video_id": "MOSE__97b0811e", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did the 3D distance between object A (large adult elephant walking to the right) and object B (small juvenile elephant walking to the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/fa212919fc2038cb1756/F0.jpg", "media/fa212919fc2038cb1756/F1.jpg", "media/fa212919fc2038cb1756/F2.jpg", "media/fa212919fc2038cb1756/F3.jpg", "media/fa212919fc2038cb1756/F4.jpg", "media/fa212919fc2038cb1756/F5.jpg", "media/fa212919fc2038cb1756/F6.jpg", "media/fa212919fc2038cb1756/F7.jpg", "media/fa212919fc2038cb1756/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1", "g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (large adult elephant walking to the right) and object B (small juvenile elephant walking to the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a4fd7df88e81cfa976449dffeab2492c3c87a2db6ea0fe4edd79c05aa97bc00"}
{"qa_id": "MOSE__97b0811e__object_travel_distance_comparison__06", "video_id": "MOSE__97b0811e", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking to the right) or B (small juvenile elephant walking to the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/fa212919fc2038cb1756/F0.jpg", "media/fa212919fc2038cb1756/F1.jpg", "media/fa212919fc2038cb1756/F2.jpg", "media/fa212919fc2038cb1756/F3.jpg", "media/fa212919fc2038cb1756/F4.jpg", "media/fa212919fc2038cb1756/F5.jpg", "media/fa212919fc2038cb1756/F6.jpg", "media/fa212919fc2038cb1756/F7.jpg", "media/fa212919fc2038cb1756/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1", "g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking to the right) or B (small juvenile elephant walking to the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2e8332e475e4f0d1bd2ea8d8e6f78f340c1e88f1756f44d887bf16576d43c13"}
{"qa_id": "MOSE__980baf58__object_world_motion_direction__00", "video_id": "MOSE__980baf58", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (orange goldfish) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/1b8c404c5776892a9087/F0.jpg", "media/1b8c404c5776892a9087/F1.jpg", "media/1b8c404c5776892a9087/F2.jpg", "media/1b8c404c5776892a9087/F3.jpg", "media/1b8c404c5776892a9087/F4.jpg", "media/1b8c404c5776892a9087/F5.jpg", "media/1b8c404c5776892a9087/F6.jpg", "media/1b8c404c5776892a9087/F7.jpg", "media/1b8c404c5776892a9087/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (orange goldfish) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c1fcd103c3acc560c5d4d824ba8e1d7626e25a68bead759d9883abdb5acb7bb"}
{"qa_id": "MOSE__980baf58__camera_object_distance_change__01", "video_id": "MOSE__980baf58", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did object A (orange goldfish) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/1b8c404c5776892a9087/F0.jpg", "media/1b8c404c5776892a9087/F1.jpg", "media/1b8c404c5776892a9087/F2.jpg", "media/1b8c404c5776892a9087/F3.jpg", "media/1b8c404c5776892a9087/F4.jpg", "media/1b8c404c5776892a9087/F5.jpg", "media/1b8c404c5776892a9087/F6.jpg", "media/1b8c404c5776892a9087/F7.jpg", "media/1b8c404c5776892a9087/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object A (orange goldfish) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "415325ddbc829b6df0e1763caeca9c0d0b13fcf7ebd4cbd866a2d981fb3e82f0"}
{"qa_id": "MOSE__980baf58__object_pair_distance_change__02", "video_id": "MOSE__980baf58", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (orange goldfish) and object B (white goldfish with long fins) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/4ba643eb7027abd285fe/F0.jpg", "media/4ba643eb7027abd285fe/F1.jpg", "media/4ba643eb7027abd285fe/F2.jpg", "media/4ba643eb7027abd285fe/F3.jpg", "media/4ba643eb7027abd285fe/F4.jpg", "media/4ba643eb7027abd285fe/F5.jpg", "media/4ba643eb7027abd285fe/F6.jpg", "media/4ba643eb7027abd285fe/F7.jpg", "media/4ba643eb7027abd285fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (orange goldfish) and object B (white goldfish with long fins) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "43eb20c3b25b99d064c21fbf5fe61dbdf1f98f0710cf05aa2134b04989b2e3ea"}
{"qa_id": "MOSE__980baf58__object_travel_distance_comparison__03", "video_id": "MOSE__980baf58", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange goldfish) or B (white goldfish with long fins)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4ba643eb7027abd285fe/F0.jpg", "media/4ba643eb7027abd285fe/F1.jpg", "media/4ba643eb7027abd285fe/F2.jpg", "media/4ba643eb7027abd285fe/F3.jpg", "media/4ba643eb7027abd285fe/F4.jpg", "media/4ba643eb7027abd285fe/F5.jpg", "media/4ba643eb7027abd285fe/F6.jpg", "media/4ba643eb7027abd285fe/F7.jpg", "media/4ba643eb7027abd285fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange goldfish) or B (white goldfish with long fins)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3f20e2a44db18ab341d4185233a67df0e3b7a596df077c703046f9e58b844795"}
{"qa_id": "MOSE__99350a74__object_pair_distance_change__00", "video_id": "MOSE__99350a74", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (white duck with grey patches and an orange beak) and object B (brown chicken with a red comb) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d09f8b2f79db5c82a6d5/F0.jpg", "media/d09f8b2f79db5c82a6d5/F1.jpg", "media/d09f8b2f79db5c82a6d5/F2.jpg", "media/d09f8b2f79db5c82a6d5/F3.jpg", "media/d09f8b2f79db5c82a6d5/F4.jpg", "media/d09f8b2f79db5c82a6d5/F5.jpg", "media/d09f8b2f79db5c82a6d5/F6.jpg", "media/d09f8b2f79db5c82a6d5/F7.jpg", "media/d09f8b2f79db5c82a6d5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (white duck with grey patches and an orange beak) and object B (brown chicken with a red comb) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7259c84d0185272d340e100d91b020dfeaefd5f04d556c2cb662c4d1783d9e7b"}
{"qa_id": "MOSE__9a48c0c9__object_world_motion_direction__00", "video_id": "MOSE__9a48c0c9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (young elephant walking on the grass) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/2cbd5bcb1c90099c8a58/F0.jpg", "media/2cbd5bcb1c90099c8a58/F1.jpg", "media/2cbd5bcb1c90099c8a58/F2.jpg", "media/2cbd5bcb1c90099c8a58/F3.jpg", "media/2cbd5bcb1c90099c8a58/F4.jpg", "media/2cbd5bcb1c90099c8a58/F5.jpg", "media/2cbd5bcb1c90099c8a58/F6.jpg", "media/2cbd5bcb1c90099c8a58/F7.jpg", "media/2cbd5bcb1c90099c8a58/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (young elephant walking on the grass) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36175f5072659a6ec1abf68696b143c475a8dee0fa3bf3ca1b6a9e9b258d51ed"}
{"qa_id": "MOSE__9a48c0c9__object_world_motion_direction__01", "video_id": "MOSE__9a48c0c9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object B (young elephant walking on the grass to the left of g2) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "left", "forward", "backward"], "frame_paths": ["media/5dfcf1c931dca0eda1a3/F0.jpg", "media/5dfcf1c931dca0eda1a3/F1.jpg", "media/5dfcf1c931dca0eda1a3/F2.jpg", "media/5dfcf1c931dca0eda1a3/F3.jpg", "media/5dfcf1c931dca0eda1a3/F4.jpg", "media/5dfcf1c931dca0eda1a3/F5.jpg", "media/5dfcf1c931dca0eda1a3/F6.jpg", "media/5dfcf1c931dca0eda1a3/F7.jpg", "media/5dfcf1c931dca0eda1a3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object B (young elephant walking on the grass to the left of g2) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e35e8db4263661017d3d4cb0334c12eebafef9f96877f74ee88fefff65ddbb8"}
{"qa_id": "MOSE__9ae447a7__object_pair_distance_change__00", "video_id": "MOSE__9ae447a7", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did the 3D distance between object A (brown duck with a lighter neck) and object B (dark brown duck) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/177873f20de89cd11fb7/F0.jpg", "media/177873f20de89cd11fb7/F1.jpg", "media/177873f20de89cd11fb7/F2.jpg", "media/177873f20de89cd11fb7/F3.jpg", "media/177873f20de89cd11fb7/F4.jpg", "media/177873f20de89cd11fb7/F5.jpg", "media/177873f20de89cd11fb7/F6.jpg", "media/177873f20de89cd11fb7/F7.jpg", "media/177873f20de89cd11fb7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (brown duck with a lighter neck) and object B (dark brown duck) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb868afc00fb88334cd4a729446c93f7c84eb8d01b8a0d33c3855ba90c3a338f"}
{"qa_id": "MOSE__9ae447a7__object_travel_distance_comparison__01", "video_id": "MOSE__9ae447a7", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown duck with a lighter neck) or B (dark brown duck)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/177873f20de89cd11fb7/F0.jpg", "media/177873f20de89cd11fb7/F1.jpg", "media/177873f20de89cd11fb7/F2.jpg", "media/177873f20de89cd11fb7/F3.jpg", "media/177873f20de89cd11fb7/F4.jpg", "media/177873f20de89cd11fb7/F5.jpg", "media/177873f20de89cd11fb7/F6.jpg", "media/177873f20de89cd11fb7/F7.jpg", "media/177873f20de89cd11fb7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown duck with a lighter neck) or B (dark brown duck)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3b6b7224c91f10cf59df5a0807e576a40a16e99a435c7f39f1b969150ebf400"}
{"qa_id": "MOSE__9b5f0f47__object_world_motion_direction__00", "video_id": "MOSE__9b5f0f47", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (orange goldfish with long flowing fins) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/b3a5c2011b1253b87183/F0.jpg", "media/b3a5c2011b1253b87183/F1.jpg", "media/b3a5c2011b1253b87183/F2.jpg", "media/b3a5c2011b1253b87183/F3.jpg", "media/b3a5c2011b1253b87183/F4.jpg", "media/b3a5c2011b1253b87183/F5.jpg", "media/b3a5c2011b1253b87183/F6.jpg", "media/b3a5c2011b1253b87183/F7.jpg", "media/b3a5c2011b1253b87183/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (orange goldfish with long flowing fins) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e58df998965d34b103f3a7a6dbc9c13675ea017a7576c0e361298ef7156a90ba"}
{"qa_id": "MOSE__9cb1ba0e__camera_translation_direction__00", "video_id": "MOSE__9cb1ba0e", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "forward", "down", "right"], "frame_paths": ["media/40f015bb571262661585/F0.jpg", "media/40f015bb571262661585/F1.jpg", "media/40f015bb571262661585/F2.jpg", "media/40f015bb571262661585/F3.jpg", "media/40f015bb571262661585/F4.jpg", "media/40f015bb571262661585/F5.jpg", "media/40f015bb571262661585/F6.jpg", "media/40f015bb571262661585/F7.jpg", "media/40f015bb571262661585/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. forward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d368a7b5cbbddc05b0248f6a02a70b508a151b410424e23367a98e8684da09f0"}
{"qa_id": "MOSE__9d299f60__camera_view_direction_change__00", "video_id": "MOSE__9d299f60", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/11cd1a76305bdd9156df/F0.jpg", "media/11cd1a76305bdd9156df/F1.jpg", "media/11cd1a76305bdd9156df/F2.jpg", "media/11cd1a76305bdd9156df/F3.jpg", "media/11cd1a76305bdd9156df/F4.jpg", "media/11cd1a76305bdd9156df/F5.jpg", "media/11cd1a76305bdd9156df/F6.jpg", "media/11cd1a76305bdd9156df/F7.jpg", "media/11cd1a76305bdd9156df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b05c59c5ff5c9031ea68746d67c85c07217775e2ecc9e157cbe29c5a6487d912"}
{"qa_id": "MOSE__9d299f60__object_pair_distance_change__01", "video_id": "MOSE__9d299f60", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did the 3D distance between object A (brown horse with a dark tail) and object B (light brown horse with a dark mane) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/0ef23e1c1c716f06e000/F0.jpg", "media/0ef23e1c1c716f06e000/F1.jpg", "media/0ef23e1c1c716f06e000/F2.jpg", "media/0ef23e1c1c716f06e000/F3.jpg", "media/0ef23e1c1c716f06e000/F4.jpg", "media/0ef23e1c1c716f06e000/F5.jpg", "media/0ef23e1c1c716f06e000/F6.jpg", "media/0ef23e1c1c716f06e000/F7.jpg", "media/0ef23e1c1c716f06e000/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (brown horse with a dark tail) and object B (light brown horse with a dark mane) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3b61d03ac520a73ad891365cedd6bf7fdc892330e290c4dd52ae964983d58348"}
{"qa_id": "MOSE__9d299f60__object_travel_distance_comparison__02", "video_id": "MOSE__9d299f60", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse with a dark tail) or B (light brown horse with a dark mane)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/0ef23e1c1c716f06e000/F0.jpg", "media/0ef23e1c1c716f06e000/F1.jpg", "media/0ef23e1c1c716f06e000/F2.jpg", "media/0ef23e1c1c716f06e000/F3.jpg", "media/0ef23e1c1c716f06e000/F4.jpg", "media/0ef23e1c1c716f06e000/F5.jpg", "media/0ef23e1c1c716f06e000/F6.jpg", "media/0ef23e1c1c716f06e000/F7.jpg", "media/0ef23e1c1c716f06e000/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse with a dark tail) or B (light brown horse with a dark mane)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7d02f45f3c647a349059698322218d051e359f7cc91a12ccafebfdc7e05ad723"}
{"qa_id": "MOSE__9d5940d2__object_pair_distance_change__00", "video_id": "MOSE__9d5940d2", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (turtle with dark brown shell on the right) and object B (turtle with lighter brown shell on the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/55a196131cf775141167/F0.jpg", "media/55a196131cf775141167/F1.jpg", "media/55a196131cf775141167/F2.jpg", "media/55a196131cf775141167/F3.jpg", "media/55a196131cf775141167/F4.jpg", "media/55a196131cf775141167/F5.jpg", "media/55a196131cf775141167/F6.jpg", "media/55a196131cf775141167/F7.jpg", "media/55a196131cf775141167/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (turtle with dark brown shell on the right) and object B (turtle with lighter brown shell on the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4cab642650a8ba047d75ab34444ccb63e93d6619f49296c9fb6420564cc1ae92"}
{"qa_id": "MOSE__9d9ae988__object_travel_distance_comparison__00", "video_id": "MOSE__9d9ae988", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark buffalo partially obscured by green bushes on the right) or B (large dark buffalo with curved horns in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/d1b6731bd6ff81dde56d/F0.jpg", "media/d1b6731bd6ff81dde56d/F1.jpg", "media/d1b6731bd6ff81dde56d/F2.jpg", "media/d1b6731bd6ff81dde56d/F3.jpg", "media/d1b6731bd6ff81dde56d/F4.jpg", "media/d1b6731bd6ff81dde56d/F5.jpg", "media/d1b6731bd6ff81dde56d/F6.jpg", "media/d1b6731bd6ff81dde56d/F7.jpg", "media/d1b6731bd6ff81dde56d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark buffalo partially obscured by green bushes on the right) or B (large dark buffalo with curved horns in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eb11595ef3e1fefefee55ee51cc8dee489a6e0463d8f008a0685fffadfbacb23"}
{"qa_id": "MOSE__9dfa2cc5__camera_view_direction_change__00", "video_id": "MOSE__9dfa2cc5", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/cf07e1dfb558be8af445/F0.jpg", "media/cf07e1dfb558be8af445/F1.jpg", "media/cf07e1dfb558be8af445/F2.jpg", "media/cf07e1dfb558be8af445/F3.jpg", "media/cf07e1dfb558be8af445/F4.jpg", "media/cf07e1dfb558be8af445/F5.jpg", "media/cf07e1dfb558be8af445/F6.jpg", "media/cf07e1dfb558be8af445/F7.jpg", "media/cf07e1dfb558be8af445/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "MOSE__9dfa2cc5__object_world_motion_direction__01", "video_id": "MOSE__9dfa2cc5", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (large adult elephant walking in the herd) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "backward", "forward"], "frame_paths": ["media/498accdfee9b4ddef22a/F0.jpg", "media/498accdfee9b4ddef22a/F1.jpg", "media/498accdfee9b4ddef22a/F2.jpg", "media/498accdfee9b4ddef22a/F3.jpg", "media/498accdfee9b4ddef22a/F4.jpg", "media/498accdfee9b4ddef22a/F5.jpg", "media/498accdfee9b4ddef22a/F6.jpg", "media/498accdfee9b4ddef22a/F7.jpg", "media/498accdfee9b4ddef22a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (large adult elephant walking in the herd) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b059095bca223fd3cf71b367ce86ca36f39efa5a809cf8ba1f6c17a2c1ad93c"}
{"qa_id": "MOSE__9dfa2cc5__camera_object_distance_change__02", "video_id": "MOSE__9dfa2cc5", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object A (large adult elephant walking in the herd) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/498accdfee9b4ddef22a/F0.jpg", "media/498accdfee9b4ddef22a/F1.jpg", "media/498accdfee9b4ddef22a/F2.jpg", "media/498accdfee9b4ddef22a/F3.jpg", "media/498accdfee9b4ddef22a/F4.jpg", "media/498accdfee9b4ddef22a/F5.jpg", "media/498accdfee9b4ddef22a/F6.jpg", "media/498accdfee9b4ddef22a/F7.jpg", "media/498accdfee9b4ddef22a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (large adult elephant walking in the herd) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4262f687bd4941142d47cb6b0692b18bb2ba8c3764239749bb44b41a0293e5c0"}
{"qa_id": "MOSE__9dfa2cc5__object_world_motion_direction__03", "video_id": "MOSE__9dfa2cc5", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object B (small juvenile elephant walking near the front) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "forward", "up", "down"], "frame_paths": ["media/61248ee0ee5c00d5f85f/F0.jpg", "media/61248ee0ee5c00d5f85f/F1.jpg", "media/61248ee0ee5c00d5f85f/F2.jpg", "media/61248ee0ee5c00d5f85f/F3.jpg", "media/61248ee0ee5c00d5f85f/F4.jpg", "media/61248ee0ee5c00d5f85f/F5.jpg", "media/61248ee0ee5c00d5f85f/F6.jpg", "media/61248ee0ee5c00d5f85f/F7.jpg", "media/61248ee0ee5c00d5f85f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object B (small juvenile elephant walking near the front) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b55a649039308cdaae439c679874640afc580f97cc45cf23f0d80873d233de8d"}
{"qa_id": "MOSE__9dfa2cc5__camera_object_distance_change__04", "video_id": "MOSE__9dfa2cc5", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object B (small juvenile elephant walking near the front) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/61248ee0ee5c00d5f85f/F0.jpg", "media/61248ee0ee5c00d5f85f/F1.jpg", "media/61248ee0ee5c00d5f85f/F2.jpg", "media/61248ee0ee5c00d5f85f/F3.jpg", "media/61248ee0ee5c00d5f85f/F4.jpg", "media/61248ee0ee5c00d5f85f/F5.jpg", "media/61248ee0ee5c00d5f85f/F6.jpg", "media/61248ee0ee5c00d5f85f/F7.jpg", "media/61248ee0ee5c00d5f85f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object B (small juvenile elephant walking near the front) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "63f77115511e84546bd47282ab704b2bbac0a19d36b4d0ecf4a3befcc8df9272"}
{"qa_id": "MOSE__9dfa2cc5__object_pair_distance_change__05", "video_id": "MOSE__9dfa2cc5", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did the 3D distance between object A (large adult elephant walking in the herd) and object B (small juvenile elephant walking near the front) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4c2bab07fa4bf0495224/F0.jpg", "media/4c2bab07fa4bf0495224/F1.jpg", "media/4c2bab07fa4bf0495224/F2.jpg", "media/4c2bab07fa4bf0495224/F3.jpg", "media/4c2bab07fa4bf0495224/F4.jpg", "media/4c2bab07fa4bf0495224/F5.jpg", "media/4c2bab07fa4bf0495224/F6.jpg", "media/4c2bab07fa4bf0495224/F7.jpg", "media/4c2bab07fa4bf0495224/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (large adult elephant walking in the herd) and object B (small juvenile elephant walking near the front) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f864477661ce997c3bff8a3b3822583f963793c2f81e5c1adcc41cd005481c0a"}
{"qa_id": "MOSE__9dfa2cc5__object_travel_distance_comparison__06", "video_id": "MOSE__9dfa2cc5", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking in the herd) or B (small juvenile elephant walking near the front)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4c2bab07fa4bf0495224/F0.jpg", "media/4c2bab07fa4bf0495224/F1.jpg", "media/4c2bab07fa4bf0495224/F2.jpg", "media/4c2bab07fa4bf0495224/F3.jpg", "media/4c2bab07fa4bf0495224/F4.jpg", "media/4c2bab07fa4bf0495224/F5.jpg", "media/4c2bab07fa4bf0495224/F6.jpg", "media/4c2bab07fa4bf0495224/F7.jpg", "media/4c2bab07fa4bf0495224/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large adult elephant walking in the herd) or B (small juvenile elephant walking near the front)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e4bf8fcd844059d75c59a09d61141fcca02e8b9b34576c82bbaab97e17d76e0"}
{"qa_id": "MOSE__a034feb2__object_pair_distance_change__00", "video_id": "MOSE__a034feb2", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did the 3D distance between object A (person in red shirt and black shorts) and object B (person in black shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/e7f8b9d7146eb5bdc96e/F0.jpg", "media/e7f8b9d7146eb5bdc96e/F1.jpg", "media/e7f8b9d7146eb5bdc96e/F2.jpg", "media/e7f8b9d7146eb5bdc96e/F3.jpg", "media/e7f8b9d7146eb5bdc96e/F4.jpg", "media/e7f8b9d7146eb5bdc96e/F5.jpg", "media/e7f8b9d7146eb5bdc96e/F6.jpg", "media/e7f8b9d7146eb5bdc96e/F7.jpg", "media/e7f8b9d7146eb5bdc96e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g7"}, "requested_object_ids": ["g1", "g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (person in red shirt and black shorts) and object B (person in black shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d5e725837010a82c9108919b2b554b26fce001e0979943252ce0eb2195fecb7"}
{"qa_id": "MOSE__a14de179__camera_translation_direction__00", "video_id": "MOSE__a14de179", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/9d9b8c0347d147f4b814/F0.jpg", "media/9d9b8c0347d147f4b814/F1.jpg", "media/9d9b8c0347d147f4b814/F2.jpg", "media/9d9b8c0347d147f4b814/F3.jpg", "media/9d9b8c0347d147f4b814/F4.jpg", "media/9d9b8c0347d147f4b814/F5.jpg", "media/9d9b8c0347d147f4b814/F6.jpg", "media/9d9b8c0347d147f4b814/F7.jpg", "media/9d9b8c0347d147f4b814/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0c577729f831a1bd18d5a2d9d8375ac87b928c6f1df695834d306e9b839a3704"}
{"qa_id": "MOSE__a25316a3__object_travel_distance_comparison__00", "video_id": "MOSE__a25316a3", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown chicken pecking at the ground) or B (brown chicken on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/da44d6fb3cfdc8962fca/F0.jpg", "media/da44d6fb3cfdc8962fca/F1.jpg", "media/da44d6fb3cfdc8962fca/F2.jpg", "media/da44d6fb3cfdc8962fca/F3.jpg", "media/da44d6fb3cfdc8962fca/F4.jpg", "media/da44d6fb3cfdc8962fca/F5.jpg", "media/da44d6fb3cfdc8962fca/F6.jpg", "media/da44d6fb3cfdc8962fca/F7.jpg", "media/da44d6fb3cfdc8962fca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown chicken pecking at the ground) or B (brown chicken on the far right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f5dc972d4f8a4ee673d4b7d5575ddbd80c90395f8dc8b9ce648cca79ca08542"}
{"qa_id": "MOSE__a3c56751__camera_object_distance_change__00", "video_id": "MOSE__a3c56751", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object A (person in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/e5a70b0010f275460b64/F0.jpg", "media/e5a70b0010f275460b64/F1.jpg", "media/e5a70b0010f275460b64/F2.jpg", "media/e5a70b0010f275460b64/F3.jpg", "media/e5a70b0010f275460b64/F4.jpg", "media/e5a70b0010f275460b64/F5.jpg", "media/e5a70b0010f275460b64/F6.jpg", "media/e5a70b0010f275460b64/F7.jpg", "media/e5a70b0010f275460b64/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person in white t-shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee5d6a3fda99b8f1feaf4e817023a08f5c3e871b9c41de2b53e880ac097c7119"}
{"qa_id": "MOSE__a4457543__object_world_motion_direction__00", "video_id": "MOSE__a4457543", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object A (large adult elephant walking on the paved road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "forward", "left", "down"], "frame_paths": ["media/39f44617e1d52cc04c29/F0.jpg", "media/39f44617e1d52cc04c29/F1.jpg", "media/39f44617e1d52cc04c29/F2.jpg", "media/39f44617e1d52cc04c29/F3.jpg", "media/39f44617e1d52cc04c29/F4.jpg", "media/39f44617e1d52cc04c29/F5.jpg", "media/39f44617e1d52cc04c29/F6.jpg", "media/39f44617e1d52cc04c29/F7.jpg", "media/39f44617e1d52cc04c29/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object A (large adult elephant walking on the paved road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b3e6004b96579cc3e994aa14dc502ce4717c041d4a0fb04a7a5c0b37b93b9c57"}
{"qa_id": "MOSE__a4457543__object_world_motion_direction__01", "video_id": "MOSE__a4457543", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object B (adult elephant walking on the dirt shoulder to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "right", "up", "backward"], "frame_paths": ["media/6781a7f90ef42fc38fda/F0.jpg", "media/6781a7f90ef42fc38fda/F1.jpg", "media/6781a7f90ef42fc38fda/F2.jpg", "media/6781a7f90ef42fc38fda/F3.jpg", "media/6781a7f90ef42fc38fda/F4.jpg", "media/6781a7f90ef42fc38fda/F5.jpg", "media/6781a7f90ef42fc38fda/F6.jpg", "media/6781a7f90ef42fc38fda/F7.jpg", "media/6781a7f90ef42fc38fda/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object B (adult elephant walking on the dirt shoulder to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0bb27233159f3a8ee228b94436b9dd9f1b311ebb4d2bb17c1813be3c728d654b"}
{"qa_id": "MOSE__a46b7b77__camera_view_direction_change__00", "video_id": "MOSE__a46b7b77", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/789270049daa1e6dc9c6/F0.jpg", "media/789270049daa1e6dc9c6/F1.jpg", "media/789270049daa1e6dc9c6/F2.jpg", "media/789270049daa1e6dc9c6/F3.jpg", "media/789270049daa1e6dc9c6/F4.jpg", "media/789270049daa1e6dc9c6/F5.jpg", "media/789270049daa1e6dc9c6/F6.jpg", "media/789270049daa1e6dc9c6/F7.jpg", "media/789270049daa1e6dc9c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4a7e894a1fab8c8ffc2737556cdacdce0495b813831fa06e0de06cb3e272b011"}
{"qa_id": "MOSE__a4fc4f17__camera_view_direction_change__00", "video_id": "MOSE__a4fc4f17", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/37317e23b1a343b3515c/F0.jpg", "media/37317e23b1a343b3515c/F1.jpg", "media/37317e23b1a343b3515c/F2.jpg", "media/37317e23b1a343b3515c/F3.jpg", "media/37317e23b1a343b3515c/F4.jpg", "media/37317e23b1a343b3515c/F5.jpg", "media/37317e23b1a343b3515c/F6.jpg", "media/37317e23b1a343b3515c/F7.jpg", "media/37317e23b1a343b3515c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "689dae4aa0b4f962ab5c50120bf67e432403813d18160a815794f01d30ce4b3c"}
{"qa_id": "MOSE__a5905593__camera_view_direction_change__00", "video_id": "MOSE__a5905593", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/3309b880a29724de085c/F0.jpg", "media/3309b880a29724de085c/F1.jpg", "media/3309b880a29724de085c/F2.jpg", "media/3309b880a29724de085c/F3.jpg", "media/3309b880a29724de085c/F4.jpg", "media/3309b880a29724de085c/F5.jpg", "media/3309b880a29724de085c/F6.jpg", "media/3309b880a29724de085c/F7.jpg", "media/3309b880a29724de085c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "MOSE__a5905593__camera_object_distance_change__01", "video_id": "MOSE__a5905593", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (blue bus with NTL branding) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/cca6f0c6014551d24a98/F0.jpg", "media/cca6f0c6014551d24a98/F1.jpg", "media/cca6f0c6014551d24a98/F2.jpg", "media/cca6f0c6014551d24a98/F3.jpg", "media/cca6f0c6014551d24a98/F4.jpg", "media/cca6f0c6014551d24a98/F5.jpg", "media/cca6f0c6014551d24a98/F6.jpg", "media/cca6f0c6014551d24a98/F7.jpg", "media/cca6f0c6014551d24a98/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (blue bus with NTL branding) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "08ae9da93ca381a39406ad44d5f9212a1ce3ab474fe5593529c2cd68166a76b3"}
{"qa_id": "MOSE__a5905593__object_pair_distance_change__02", "video_id": "MOSE__a5905593", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (blue bus with NTL branding) and object B (white and red double-decker bus) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/1d7feee4a93d68012c54/F0.jpg", "media/1d7feee4a93d68012c54/F1.jpg", "media/1d7feee4a93d68012c54/F2.jpg", "media/1d7feee4a93d68012c54/F3.jpg", "media/1d7feee4a93d68012c54/F4.jpg", "media/1d7feee4a93d68012c54/F5.jpg", "media/1d7feee4a93d68012c54/F6.jpg", "media/1d7feee4a93d68012c54/F7.jpg", "media/1d7feee4a93d68012c54/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (blue bus with NTL branding) and object B (white and red double-decker bus) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db5118e0893c6d6f0ea076a63de707aef0fb29fd15a5ea68e77b7bbc9f1ba86e"}
{"qa_id": "MOSE__a5905593__object_travel_distance_comparison__03", "video_id": "MOSE__a5905593", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (blue bus with NTL branding) or B (white and red double-decker bus)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/1d7feee4a93d68012c54/F0.jpg", "media/1d7feee4a93d68012c54/F1.jpg", "media/1d7feee4a93d68012c54/F2.jpg", "media/1d7feee4a93d68012c54/F3.jpg", "media/1d7feee4a93d68012c54/F4.jpg", "media/1d7feee4a93d68012c54/F5.jpg", "media/1d7feee4a93d68012c54/F6.jpg", "media/1d7feee4a93d68012c54/F7.jpg", "media/1d7feee4a93d68012c54/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (blue bus with NTL branding) or B (white and red double-decker bus)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "45f65a601f22c30b90287fa18917c88053702b91840c4ab60bf52578ac6b0482"}
{"qa_id": "MOSE__a5f70ce7__camera_object_distance_change__00", "video_id": "MOSE__a5f70ce7", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did object B (dark-shelled turtle with a patterned head and extended front flippers) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/97eee2ec4617bd92c83e/F0.jpg", "media/97eee2ec4617bd92c83e/F1.jpg", "media/97eee2ec4617bd92c83e/F2.jpg", "media/97eee2ec4617bd92c83e/F3.jpg", "media/97eee2ec4617bd92c83e/F4.jpg", "media/97eee2ec4617bd92c83e/F5.jpg", "media/97eee2ec4617bd92c83e/F6.jpg", "media/97eee2ec4617bd92c83e/F7.jpg", "media/97eee2ec4617bd92c83e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object B (dark-shelled turtle with a patterned head and extended front flippers) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3efc897d18176255ca4c342973c08c0d1305941397462cb53e265c600977e58d"}
{"qa_id": "MOSE__a5f70ce7__object_pair_distance_change__01", "video_id": "MOSE__a5f70ce7", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did the 3D distance between object A (dark-shelled turtle with a visible head and front flippers) and object B (dark-shelled turtle with a patterned head and extended front flippers) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ee5878e2d41247ab61ef/F0.jpg", "media/ee5878e2d41247ab61ef/F1.jpg", "media/ee5878e2d41247ab61ef/F2.jpg", "media/ee5878e2d41247ab61ef/F3.jpg", "media/ee5878e2d41247ab61ef/F4.jpg", "media/ee5878e2d41247ab61ef/F5.jpg", "media/ee5878e2d41247ab61ef/F6.jpg", "media/ee5878e2d41247ab61ef/F7.jpg", "media/ee5878e2d41247ab61ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (dark-shelled turtle with a visible head and front flippers) and object B (dark-shelled turtle with a patterned head and extended front flippers) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96185d6b36b11c061a6b7e0bf5136cc4a4cf43a81f1d69c6b064d22b77434a00"}
{"qa_id": "MOSE__a5f70ce7__object_travel_distance_comparison__02", "video_id": "MOSE__a5f70ce7", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark-shelled turtle with a visible head and front flippers) or B (dark-shelled turtle with a patterned head and extended front flippers)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/ee5878e2d41247ab61ef/F0.jpg", "media/ee5878e2d41247ab61ef/F1.jpg", "media/ee5878e2d41247ab61ef/F2.jpg", "media/ee5878e2d41247ab61ef/F3.jpg", "media/ee5878e2d41247ab61ef/F4.jpg", "media/ee5878e2d41247ab61ef/F5.jpg", "media/ee5878e2d41247ab61ef/F6.jpg", "media/ee5878e2d41247ab61ef/F7.jpg", "media/ee5878e2d41247ab61ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark-shelled turtle with a visible head and front flippers) or B (dark-shelled turtle with a patterned head and extended front flippers)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "54fd02219c648d7e96613240537eea5bf9eff5d0a0bf1709862bf018f9c90b23"}
{"qa_id": "MOSE__a7eb6616__object_world_motion_direction__00", "video_id": "MOSE__a7eb6616", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (dark brown bull with horns) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "down", "up", "forward"], "frame_paths": ["media/5ac87e1eed5645eb60f9/F0.jpg", "media/5ac87e1eed5645eb60f9/F1.jpg", "media/5ac87e1eed5645eb60f9/F2.jpg", "media/5ac87e1eed5645eb60f9/F3.jpg", "media/5ac87e1eed5645eb60f9/F4.jpg", "media/5ac87e1eed5645eb60f9/F5.jpg", "media/5ac87e1eed5645eb60f9/F6.jpg", "media/5ac87e1eed5645eb60f9/F7.jpg", "media/5ac87e1eed5645eb60f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (dark brown bull with horns) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c589250b621c8f07cb80736ca10c60989e9a088e1f8845191a6feded85b577df"}
{"qa_id": "MOSE__a810399b__camera_translation_direction__00", "video_id": "MOSE__a810399b", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/4ceb16c047bbbae5657f/F0.jpg", "media/4ceb16c047bbbae5657f/F1.jpg", "media/4ceb16c047bbbae5657f/F2.jpg", "media/4ceb16c047bbbae5657f/F3.jpg", "media/4ceb16c047bbbae5657f/F4.jpg", "media/4ceb16c047bbbae5657f/F5.jpg", "media/4ceb16c047bbbae5657f/F6.jpg", "media/4ceb16c047bbbae5657f/F7.jpg", "media/4ceb16c047bbbae5657f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "46a10855696180925af687dc85439412b11db77fa1becc4b85ede3e5494bb168"}
{"qa_id": "MOSE__a810399b__object_world_motion_direction__01", "video_id": "MOSE__a810399b", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in light grey t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/f3664bd526a90b270592/F0.jpg", "media/f3664bd526a90b270592/F1.jpg", "media/f3664bd526a90b270592/F2.jpg", "media/f3664bd526a90b270592/F3.jpg", "media/f3664bd526a90b270592/F4.jpg", "media/f3664bd526a90b270592/F5.jpg", "media/f3664bd526a90b270592/F6.jpg", "media/f3664bd526a90b270592/F7.jpg", "media/f3664bd526a90b270592/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in light grey t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0264f15a84e9c0ec5d21e3faafbdb82fa2993a295989ab6f9c6e4021161f6880"}
{"qa_id": "MOSE__a810399b__camera_object_distance_change__02", "video_id": "MOSE__a810399b", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (person in light grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f3664bd526a90b270592/F0.jpg", "media/f3664bd526a90b270592/F1.jpg", "media/f3664bd526a90b270592/F2.jpg", "media/f3664bd526a90b270592/F3.jpg", "media/f3664bd526a90b270592/F4.jpg", "media/f3664bd526a90b270592/F5.jpg", "media/f3664bd526a90b270592/F6.jpg", "media/f3664bd526a90b270592/F7.jpg", "media/f3664bd526a90b270592/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (person in light grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "46036e69deebc98bf393d752f844b4d03479732f8973bd7e4fcff04ff4224bdb"}
{"qa_id": "MOSE__a810399b__object_pair_distance_change__03", "video_id": "MOSE__a810399b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (person in light grey t-shirt) and object B (person in white t-shirt and black pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/16eca65fc4f7061eb255/F0.jpg", "media/16eca65fc4f7061eb255/F1.jpg", "media/16eca65fc4f7061eb255/F2.jpg", "media/16eca65fc4f7061eb255/F3.jpg", "media/16eca65fc4f7061eb255/F4.jpg", "media/16eca65fc4f7061eb255/F5.jpg", "media/16eca65fc4f7061eb255/F6.jpg", "media/16eca65fc4f7061eb255/F7.jpg", "media/16eca65fc4f7061eb255/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in light grey t-shirt) and object B (person in white t-shirt and black pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cf2253a01ee39d9151824cf322122baaec763d8b7941ef5435db27fe039198bf"}
{"qa_id": "MOSE__a810399b__object_travel_distance_comparison__04", "video_id": "MOSE__a810399b", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in light grey t-shirt) or B (person in white t-shirt and black pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/16eca65fc4f7061eb255/F0.jpg", "media/16eca65fc4f7061eb255/F1.jpg", "media/16eca65fc4f7061eb255/F2.jpg", "media/16eca65fc4f7061eb255/F3.jpg", "media/16eca65fc4f7061eb255/F4.jpg", "media/16eca65fc4f7061eb255/F5.jpg", "media/16eca65fc4f7061eb255/F6.jpg", "media/16eca65fc4f7061eb255/F7.jpg", "media/16eca65fc4f7061eb255/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in light grey t-shirt) or B (person in white t-shirt and black pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c0397cb768235be06c2305b590dfa361ae5d616c10d0002917044d3433a77097"}
{"qa_id": "MOSE__a86f84ee__object_world_motion_direction__00", "video_id": "MOSE__a86f84ee", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange lizard) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/33d9d0e61def7fa05742/F0.jpg", "media/33d9d0e61def7fa05742/F1.jpg", "media/33d9d0e61def7fa05742/F2.jpg", "media/33d9d0e61def7fa05742/F3.jpg", "media/33d9d0e61def7fa05742/F4.jpg", "media/33d9d0e61def7fa05742/F5.jpg", "media/33d9d0e61def7fa05742/F6.jpg", "media/33d9d0e61def7fa05742/F7.jpg", "media/33d9d0e61def7fa05742/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange lizard) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a15b5f8de0b3fcb43ea0382d9f98465bbd894f5de041e98f2a328874b903d59b"}
{"qa_id": "MOSE__a86f84ee__object_pair_distance_change__01", "video_id": "MOSE__a86f84ee", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F6 to F8, did the 3D distance between object A (orange lizard) and object B (grey lizard) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/5e7c5d1ce46edd4eb362/F0.jpg", "media/5e7c5d1ce46edd4eb362/F1.jpg", "media/5e7c5d1ce46edd4eb362/F2.jpg", "media/5e7c5d1ce46edd4eb362/F3.jpg", "media/5e7c5d1ce46edd4eb362/F4.jpg", "media/5e7c5d1ce46edd4eb362/F5.jpg", "media/5e7c5d1ce46edd4eb362/F6.jpg", "media/5e7c5d1ce46edd4eb362/F7.jpg", "media/5e7c5d1ce46edd4eb362/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (orange lizard) and object B (grey lizard) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "27fff9214369093c9eb55d5bbf9632afff225802c84fe54917f18cb67dcfe302"}
{"qa_id": "MOSE__a92390a0__object_travel_distance_comparison__00", "video_id": "MOSE__a92390a0", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with head down grazing on dry leaves) or B (sheep standing in the background facing left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/ae613aaac883008f0316/F0.jpg", "media/ae613aaac883008f0316/F1.jpg", "media/ae613aaac883008f0316/F2.jpg", "media/ae613aaac883008f0316/F3.jpg", "media/ae613aaac883008f0316/F4.jpg", "media/ae613aaac883008f0316/F5.jpg", "media/ae613aaac883008f0316/F6.jpg", "media/ae613aaac883008f0316/F7.jpg", "media/ae613aaac883008f0316/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with head down grazing on dry leaves) or B (sheep standing in the background facing left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "544e32bfe02aa10dcf73867acc02681a279a8c5b2061b7c669e3e1877fde573a"}
{"qa_id": "MOSE__a946317a__object_pair_distance_change__00", "video_id": "MOSE__a946317a", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did the 3D distance between object A (dark grey bird with head down in the water) and object B (grey and white bird standing in the grass) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ef104b96ea2b292690dd/F0.jpg", "media/ef104b96ea2b292690dd/F1.jpg", "media/ef104b96ea2b292690dd/F2.jpg", "media/ef104b96ea2b292690dd/F3.jpg", "media/ef104b96ea2b292690dd/F4.jpg", "media/ef104b96ea2b292690dd/F5.jpg", "media/ef104b96ea2b292690dd/F6.jpg", "media/ef104b96ea2b292690dd/F7.jpg", "media/ef104b96ea2b292690dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (dark grey bird with head down in the water) and object B (grey and white bird standing in the grass) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3d1737a9a76157ce1ebdd2ca0ff70b5c242f293ef7f552793f867e8e631273e"}
{"qa_id": "MOSE__a988ec74__camera_translation_direction__00", "video_id": "MOSE__a988ec74", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "left", "down", "forward"], "frame_paths": ["media/0a5f21bd99bb42ed79b8/F0.jpg", "media/0a5f21bd99bb42ed79b8/F1.jpg", "media/0a5f21bd99bb42ed79b8/F2.jpg", "media/0a5f21bd99bb42ed79b8/F3.jpg", "media/0a5f21bd99bb42ed79b8/F4.jpg", "media/0a5f21bd99bb42ed79b8/F5.jpg", "media/0a5f21bd99bb42ed79b8/F6.jpg", "media/0a5f21bd99bb42ed79b8/F7.jpg", "media/0a5f21bd99bb42ed79b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d11c9aa8cd2c20f14fd64efc68d3ecd472292faaa1166ad1142251990b73cdbd"}
{"qa_id": "MOSE__aab5ed11__camera_object_distance_change__00", "video_id": "MOSE__aab5ed11", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F7, did object A (woman in white dress roller skating) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b09f3a158ee75e19760c/F0.jpg", "media/b09f3a158ee75e19760c/F1.jpg", "media/b09f3a158ee75e19760c/F2.jpg", "media/b09f3a158ee75e19760c/F3.jpg", "media/b09f3a158ee75e19760c/F4.jpg", "media/b09f3a158ee75e19760c/F5.jpg", "media/b09f3a158ee75e19760c/F6.jpg", "media/b09f3a158ee75e19760c/F7.jpg", "media/b09f3a158ee75e19760c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, did object A (woman in white dress roller skating) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3fef1fd7a442d94e46ab76d34c4a9f21e1d8e75f0f37f0e833bf2540d291a18a"}
{"qa_id": "MOSE__aac33bff__camera_translation_direction__00", "video_id": "MOSE__aac33bff", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "up", "backward"], "frame_paths": ["media/d6e70347ba8e935a0fdd/F0.jpg", "media/d6e70347ba8e935a0fdd/F1.jpg", "media/d6e70347ba8e935a0fdd/F2.jpg", "media/d6e70347ba8e935a0fdd/F3.jpg", "media/d6e70347ba8e935a0fdd/F4.jpg", "media/d6e70347ba8e935a0fdd/F5.jpg", "media/d6e70347ba8e935a0fdd/F6.jpg", "media/d6e70347ba8e935a0fdd/F7.jpg", "media/d6e70347ba8e935a0fdd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "909d2bebfaf4d1b24fa199acebbf47b7ee7e277c82c310a95c0de66c82cbd1f0"}
{"qa_id": "MOSE__aae923b8__object_world_motion_direction__01", "video_id": "MOSE__aae923b8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (white China Eastern airplane fuselage) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/859479a33eea08c96b8e/F0.jpg", "media/859479a33eea08c96b8e/F1.jpg", "media/859479a33eea08c96b8e/F2.jpg", "media/859479a33eea08c96b8e/F3.jpg", "media/859479a33eea08c96b8e/F4.jpg", "media/859479a33eea08c96b8e/F5.jpg", "media/859479a33eea08c96b8e/F6.jpg", "media/859479a33eea08c96b8e/F7.jpg", "media/859479a33eea08c96b8e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (white China Eastern airplane fuselage) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b26be7a1724e636f31e585ed09967c79f401fb9ad804119c9aa3964440802b5c"}
{"qa_id": "MOSE__ac32a655__object_world_motion_direction__00", "video_id": "MOSE__ac32a655", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown chicken with orange legs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "right", "up", "backward"], "frame_paths": ["media/c6a7f79c3e95433331fc/F0.jpg", "media/c6a7f79c3e95433331fc/F1.jpg", "media/c6a7f79c3e95433331fc/F2.jpg", "media/c6a7f79c3e95433331fc/F3.jpg", "media/c6a7f79c3e95433331fc/F4.jpg", "media/c6a7f79c3e95433331fc/F5.jpg", "media/c6a7f79c3e95433331fc/F6.jpg", "media/c6a7f79c3e95433331fc/F7.jpg", "media/c6a7f79c3e95433331fc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown chicken with orange legs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "29622725c1ef6d2a586eec9135f189fabcb28598348ffe6f0f082c2be071b568"}
{"qa_id": "MOSE__ac32a655__object_pair_distance_change__01", "video_id": "MOSE__ac32a655", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (light brown chicken with orange legs) and object B (dark brown chicken with red comb) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/e346c4d71db2db554997/F0.jpg", "media/e346c4d71db2db554997/F1.jpg", "media/e346c4d71db2db554997/F2.jpg", "media/e346c4d71db2db554997/F3.jpg", "media/e346c4d71db2db554997/F4.jpg", "media/e346c4d71db2db554997/F5.jpg", "media/e346c4d71db2db554997/F6.jpg", "media/e346c4d71db2db554997/F7.jpg", "media/e346c4d71db2db554997/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (light brown chicken with orange legs) and object B (dark brown chicken with red comb) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5562e46e88cd9fb847b96533dbbf8307cf21e07a716493feb6db5351e53d615d"}
{"qa_id": "MOSE__ae043406__object_pair_distance_change__00", "video_id": "MOSE__ae043406", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (person in blue uniform holding a panda) and object B (panda cub on the ground) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/6dad78285fe4f91ebbc8/F0.jpg", "media/6dad78285fe4f91ebbc8/F1.jpg", "media/6dad78285fe4f91ebbc8/F2.jpg", "media/6dad78285fe4f91ebbc8/F3.jpg", "media/6dad78285fe4f91ebbc8/F4.jpg", "media/6dad78285fe4f91ebbc8/F5.jpg", "media/6dad78285fe4f91ebbc8/F6.jpg", "media/6dad78285fe4f91ebbc8/F7.jpg", "media/6dad78285fe4f91ebbc8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in blue uniform holding a panda) and object B (panda cub on the ground) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4dd378699f0a876b03cfe40ede024d7e28baceb47bd8198adf393bba172e15cc"}
{"qa_id": "MOSE__ae043406__object_travel_distance_comparison__01", "video_id": "MOSE__ae043406", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue uniform holding a panda) or B (panda cub on the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/6dad78285fe4f91ebbc8/F0.jpg", "media/6dad78285fe4f91ebbc8/F1.jpg", "media/6dad78285fe4f91ebbc8/F2.jpg", "media/6dad78285fe4f91ebbc8/F3.jpg", "media/6dad78285fe4f91ebbc8/F4.jpg", "media/6dad78285fe4f91ebbc8/F5.jpg", "media/6dad78285fe4f91ebbc8/F6.jpg", "media/6dad78285fe4f91ebbc8/F7.jpg", "media/6dad78285fe4f91ebbc8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue uniform holding a panda) or B (panda cub on the ground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73443ee0a56310f0662c8d36cd869aa57fd8b28b80ceee516809c44a9f5b5a03"}
{"qa_id": "MOSE__aef0d049__object_travel_distance_comparison__00", "video_id": "MOSE__aef0d049", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in grey striped shirt and black shorts) or B (double-decker bus with red, purple, and white livery)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/80e2f02070024cc77b45/F0.jpg", "media/80e2f02070024cc77b45/F1.jpg", "media/80e2f02070024cc77b45/F2.jpg", "media/80e2f02070024cc77b45/F3.jpg", "media/80e2f02070024cc77b45/F4.jpg", "media/80e2f02070024cc77b45/F5.jpg", "media/80e2f02070024cc77b45/F6.jpg", "media/80e2f02070024cc77b45/F7.jpg", "media/80e2f02070024cc77b45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in grey striped shirt and black shorts) or B (double-decker bus with red, purple, and white livery)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9597577e454321e1dd3593b75b8ba2789456c7904a1aedafecb9337894236419"}
{"qa_id": "MOSE__b0d09e6a__camera_translation_direction__00", "video_id": "MOSE__b0d09e6a", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "up", "backward", "down"], "frame_paths": ["media/7b03014542fede822973/F0.jpg", "media/7b03014542fede822973/F1.jpg", "media/7b03014542fede822973/F2.jpg", "media/7b03014542fede822973/F3.jpg", "media/7b03014542fede822973/F4.jpg", "media/7b03014542fede822973/F5.jpg", "media/7b03014542fede822973/F6.jpg", "media/7b03014542fede822973/F7.jpg", "media/7b03014542fede822973/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "03101c53605d0a61366aea2d099d1471562a23fe94e8b1826486ce7e01a3d752"}
{"qa_id": "MOSE__b0f7e75c__object_pair_distance_change__00", "video_id": "MOSE__b0f7e75c", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (small elephant) and object B (large elephant) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/13deeca8ecafb4331e0b/F0.jpg", "media/13deeca8ecafb4331e0b/F1.jpg", "media/13deeca8ecafb4331e0b/F2.jpg", "media/13deeca8ecafb4331e0b/F3.jpg", "media/13deeca8ecafb4331e0b/F4.jpg", "media/13deeca8ecafb4331e0b/F5.jpg", "media/13deeca8ecafb4331e0b/F6.jpg", "media/13deeca8ecafb4331e0b/F7.jpg", "media/13deeca8ecafb4331e0b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (small elephant) and object B (large elephant) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c0651c30450b18d606ee8d84392735c79fab8e144d3ba39a91907b223e0219e7"}
{"qa_id": "MOSE__b12701b7__camera_translation_direction__00", "video_id": "MOSE__b12701b7", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "right", "backward"], "frame_paths": ["media/c0eceece5e983f6ffbdb/F0.jpg", "media/c0eceece5e983f6ffbdb/F1.jpg", "media/c0eceece5e983f6ffbdb/F2.jpg", "media/c0eceece5e983f6ffbdb/F3.jpg", "media/c0eceece5e983f6ffbdb/F4.jpg", "media/c0eceece5e983f6ffbdb/F5.jpg", "media/c0eceece5e983f6ffbdb/F6.jpg", "media/c0eceece5e983f6ffbdb/F7.jpg", "media/c0eceece5e983f6ffbdb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31e1cc4e4a223b634787c2dd842f262d21ada5513f38406649586590362aadca"}
{"qa_id": "MOSE__b12701b7__camera_view_direction_change__01", "video_id": "MOSE__b12701b7", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/c0eceece5e983f6ffbdb/F0.jpg", "media/c0eceece5e983f6ffbdb/F1.jpg", "media/c0eceece5e983f6ffbdb/F2.jpg", "media/c0eceece5e983f6ffbdb/F3.jpg", "media/c0eceece5e983f6ffbdb/F4.jpg", "media/c0eceece5e983f6ffbdb/F5.jpg", "media/c0eceece5e983f6ffbdb/F6.jpg", "media/c0eceece5e983f6ffbdb/F7.jpg", "media/c0eceece5e983f6ffbdb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f589f53c223b74c5e5c9cf39ec7e8080951b0e74fc3f21d2bdc902bf6f6ecb1"}
{"qa_id": "MOSE__b12701b7__camera_object_distance_change__02", "video_id": "MOSE__b12701b7", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object B (small orange kitten in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/7380b76e4d335b54da21/F0.jpg", "media/7380b76e4d335b54da21/F1.jpg", "media/7380b76e4d335b54da21/F2.jpg", "media/7380b76e4d335b54da21/F3.jpg", "media/7380b76e4d335b54da21/F4.jpg", "media/7380b76e4d335b54da21/F5.jpg", "media/7380b76e4d335b54da21/F6.jpg", "media/7380b76e4d335b54da21/F7.jpg", "media/7380b76e4d335b54da21/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object B (small orange kitten in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82bddffd3bc36ce1f2bf6b0ef9301e902ca237cf0a68616491d0c3ef4a935215"}
{"qa_id": "MOSE__b12701b7__object_pair_distance_change__03", "video_id": "MOSE__b12701b7", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did the 3D distance between object A (small orange kitten in the background) and object B (small orange kitten in the foreground) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/fa76c8772efd40726616/F0.jpg", "media/fa76c8772efd40726616/F1.jpg", "media/fa76c8772efd40726616/F2.jpg", "media/fa76c8772efd40726616/F3.jpg", "media/fa76c8772efd40726616/F4.jpg", "media/fa76c8772efd40726616/F5.jpg", "media/fa76c8772efd40726616/F6.jpg", "media/fa76c8772efd40726616/F7.jpg", "media/fa76c8772efd40726616/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (small orange kitten in the background) and object B (small orange kitten in the foreground) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82c5a155d732f292ae8ba6e083e1af783fb6f7e590812d4111c77545c314bea0"}
{"qa_id": "MOSE__b12701b7__object_travel_distance_comparison__04", "video_id": "MOSE__b12701b7", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small orange kitten in the background) or B (small orange kitten in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/fa76c8772efd40726616/F0.jpg", "media/fa76c8772efd40726616/F1.jpg", "media/fa76c8772efd40726616/F2.jpg", "media/fa76c8772efd40726616/F3.jpg", "media/fa76c8772efd40726616/F4.jpg", "media/fa76c8772efd40726616/F5.jpg", "media/fa76c8772efd40726616/F6.jpg", "media/fa76c8772efd40726616/F7.jpg", "media/fa76c8772efd40726616/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small orange kitten in the background) or B (small orange kitten in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a07e658746049fccc19d4fc9e2881623168678ed7e3c9c03da024cb0078f4c4"}
{"qa_id": "MOSE__b30744df__object_world_motion_direction__00", "video_id": "MOSE__b30744df", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (small elephant lying down in the foreground) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "down", "right", "up"], "frame_paths": ["media/dda9538f1a6db002dbdf/F0.jpg", "media/dda9538f1a6db002dbdf/F1.jpg", "media/dda9538f1a6db002dbdf/F2.jpg", "media/dda9538f1a6db002dbdf/F3.jpg", "media/dda9538f1a6db002dbdf/F4.jpg", "media/dda9538f1a6db002dbdf/F5.jpg", "media/dda9538f1a6db002dbdf/F6.jpg", "media/dda9538f1a6db002dbdf/F7.jpg", "media/dda9538f1a6db002dbdf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (small elephant lying down in the foreground) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2b9198541c79163cfe4236a50d3a69e3c8bb44776053ec23bfab5df0333fa854"}
{"qa_id": "MOSE__b30744df__camera_object_distance_change__01", "video_id": "MOSE__b30744df", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object B (small elephant standing in the background on the right) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9a678a637cb02f185cd6/F0.jpg", "media/9a678a637cb02f185cd6/F1.jpg", "media/9a678a637cb02f185cd6/F2.jpg", "media/9a678a637cb02f185cd6/F3.jpg", "media/9a678a637cb02f185cd6/F4.jpg", "media/9a678a637cb02f185cd6/F5.jpg", "media/9a678a637cb02f185cd6/F6.jpg", "media/9a678a637cb02f185cd6/F7.jpg", "media/9a678a637cb02f185cd6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object B (small elephant standing in the background on the right) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "340c75f3d55c36d090f157a7e07ab926b7f8c316d710b1b14ae9019e0006914f"}
{"qa_id": "MOSE__b30744df__object_pair_distance_change__02", "video_id": "MOSE__b30744df", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did the 3D distance between object A (small elephant lying down in the foreground) and object B (small elephant standing in the background on the right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/038d966ed3a4e7488d7c/F0.jpg", "media/038d966ed3a4e7488d7c/F1.jpg", "media/038d966ed3a4e7488d7c/F2.jpg", "media/038d966ed3a4e7488d7c/F3.jpg", "media/038d966ed3a4e7488d7c/F4.jpg", "media/038d966ed3a4e7488d7c/F5.jpg", "media/038d966ed3a4e7488d7c/F6.jpg", "media/038d966ed3a4e7488d7c/F7.jpg", "media/038d966ed3a4e7488d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did the 3D distance between object A (small elephant lying down in the foreground) and object B (small elephant standing in the background on the right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34cc81b8de0cadb0ffea6f94c8f1bed3f2d5a02269519cc75e0ef9c50931cc61"}
{"qa_id": "MOSE__b30744df__object_travel_distance_comparison__03", "video_id": "MOSE__b30744df", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small elephant lying down in the foreground) or B (small elephant standing in the background on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/038d966ed3a4e7488d7c/F0.jpg", "media/038d966ed3a4e7488d7c/F1.jpg", "media/038d966ed3a4e7488d7c/F2.jpg", "media/038d966ed3a4e7488d7c/F3.jpg", "media/038d966ed3a4e7488d7c/F4.jpg", "media/038d966ed3a4e7488d7c/F5.jpg", "media/038d966ed3a4e7488d7c/F6.jpg", "media/038d966ed3a4e7488d7c/F7.jpg", "media/038d966ed3a4e7488d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small elephant lying down in the foreground) or B (small elephant standing in the background on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eac5ba9ad7f1932a392757bd578eb752655641e32f7059d633355304d84ab891"}
{"qa_id": "MOSE__b394847f__camera_translation_direction__00", "video_id": "MOSE__b394847f", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "right", "down", "left"], "frame_paths": ["media/65e775d959118e5dd687/F0.jpg", "media/65e775d959118e5dd687/F1.jpg", "media/65e775d959118e5dd687/F2.jpg", "media/65e775d959118e5dd687/F3.jpg", "media/65e775d959118e5dd687/F4.jpg", "media/65e775d959118e5dd687/F5.jpg", "media/65e775d959118e5dd687/F6.jpg", "media/65e775d959118e5dd687/F7.jpg", "media/65e775d959118e5dd687/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bcde9014174fc84d61ff3ff78dcb035f87c3f824405a09aac652ea94e5a8f570"}
{"qa_id": "MOSE__b3b51117__object_pair_distance_change__00", "video_id": "MOSE__b3b51117", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (light brown cow) and object B (dark brown cow) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7ed3773a37efac1afa5b/F0.jpg", "media/7ed3773a37efac1afa5b/F1.jpg", "media/7ed3773a37efac1afa5b/F2.jpg", "media/7ed3773a37efac1afa5b/F3.jpg", "media/7ed3773a37efac1afa5b/F4.jpg", "media/7ed3773a37efac1afa5b/F5.jpg", "media/7ed3773a37efac1afa5b/F6.jpg", "media/7ed3773a37efac1afa5b/F7.jpg", "media/7ed3773a37efac1afa5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (light brown cow) and object B (dark brown cow) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "266c32110cc31630efdbcc3910a5c11d748d51d92889c28af34b5d26a5a47cfa"}
{"qa_id": "MOSE__b6e2c041__object_travel_distance_comparison__00", "video_id": "MOSE__b6e2c041", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dog in pink sweater) or B (dog in dark patterned sweater)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/9bf1ff059018c6151099/F0.jpg", "media/9bf1ff059018c6151099/F1.jpg", "media/9bf1ff059018c6151099/F2.jpg", "media/9bf1ff059018c6151099/F3.jpg", "media/9bf1ff059018c6151099/F4.jpg", "media/9bf1ff059018c6151099/F5.jpg", "media/9bf1ff059018c6151099/F6.jpg", "media/9bf1ff059018c6151099/F7.jpg", "media/9bf1ff059018c6151099/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dog in pink sweater) or B (dog in dark patterned sweater)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "590c77b891df571ae0dbfc800f931a8e3d8d2a11ecbb0ed02818a4c932bfb533"}
{"qa_id": "MOSE__b6f03bd9__object_travel_distance_comparison__00", "video_id": "MOSE__b6f03bd9", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tiger with orange and black stripes) or B (tiger with orange and black stripes)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/3034d8e83ce5dfd216fb/F0.jpg", "media/3034d8e83ce5dfd216fb/F1.jpg", "media/3034d8e83ce5dfd216fb/F2.jpg", "media/3034d8e83ce5dfd216fb/F3.jpg", "media/3034d8e83ce5dfd216fb/F4.jpg", "media/3034d8e83ce5dfd216fb/F5.jpg", "media/3034d8e83ce5dfd216fb/F6.jpg", "media/3034d8e83ce5dfd216fb/F7.jpg", "media/3034d8e83ce5dfd216fb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tiger with orange and black stripes) or B (tiger with orange and black stripes)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "371394d7d20d6fcebdaa59dbc824c4f97e25b99433c294ef208a9f5dfdffaee4"}
{"qa_id": "MOSE__b7fb59c9__object_world_motion_direction__00", "video_id": "MOSE__b7fb59c9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (grey goose with black tail feathers) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/68460a0278cba4f804e4/F0.jpg", "media/68460a0278cba4f804e4/F1.jpg", "media/68460a0278cba4f804e4/F2.jpg", "media/68460a0278cba4f804e4/F3.jpg", "media/68460a0278cba4f804e4/F4.jpg", "media/68460a0278cba4f804e4/F5.jpg", "media/68460a0278cba4f804e4/F6.jpg", "media/68460a0278cba4f804e4/F7.jpg", "media/68460a0278cba4f804e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (grey goose with black tail feathers) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d2f32b8566a61d21e268478b3c9a7f919cce25ae48a046e5dff6f0c8171f8d4"}
{"qa_id": "MOSE__b7fb59c9__object_pair_distance_change__01", "video_id": "MOSE__b7fb59c9", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (grey goose with orange legs) and object B (grey goose with black tail feathers) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/c95d31a2dbf07efd6e39/F0.jpg", "media/c95d31a2dbf07efd6e39/F1.jpg", "media/c95d31a2dbf07efd6e39/F2.jpg", "media/c95d31a2dbf07efd6e39/F3.jpg", "media/c95d31a2dbf07efd6e39/F4.jpg", "media/c95d31a2dbf07efd6e39/F5.jpg", "media/c95d31a2dbf07efd6e39/F6.jpg", "media/c95d31a2dbf07efd6e39/F7.jpg", "media/c95d31a2dbf07efd6e39/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (grey goose with orange legs) and object B (grey goose with black tail feathers) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee23e0035d4587b6164587f0eb490d95da6ff3e387bab08ea6f6ede4aacb294b"}
{"qa_id": "MOSE__b7fb59c9__object_travel_distance_comparison__02", "video_id": "MOSE__b7fb59c9", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (grey goose with orange legs) or B (grey goose with black tail feathers)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/c95d31a2dbf07efd6e39/F0.jpg", "media/c95d31a2dbf07efd6e39/F1.jpg", "media/c95d31a2dbf07efd6e39/F2.jpg", "media/c95d31a2dbf07efd6e39/F3.jpg", "media/c95d31a2dbf07efd6e39/F4.jpg", "media/c95d31a2dbf07efd6e39/F5.jpg", "media/c95d31a2dbf07efd6e39/F6.jpg", "media/c95d31a2dbf07efd6e39/F7.jpg", "media/c95d31a2dbf07efd6e39/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (grey goose with orange legs) or B (grey goose with black tail feathers)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d832b84d0a5efe820b14400c0aa12d2d036cca92952cf3a1e1d1f509eb71d262"}
{"qa_id": "MOSE__b826d642__object_world_motion_direction__00", "video_id": "MOSE__b826d642", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown duck with orange beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "right", "down"], "frame_paths": ["media/63e09c7cae381201014e/F0.jpg", "media/63e09c7cae381201014e/F1.jpg", "media/63e09c7cae381201014e/F2.jpg", "media/63e09c7cae381201014e/F3.jpg", "media/63e09c7cae381201014e/F4.jpg", "media/63e09c7cae381201014e/F5.jpg", "media/63e09c7cae381201014e/F6.jpg", "media/63e09c7cae381201014e/F7.jpg", "media/63e09c7cae381201014e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown duck with orange beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2db90b6c0883506ecb986e4e2a4ce2f099c00aa63875e243d6318a672bdd81a6"}
{"qa_id": "MOSE__b826d642__object_world_motion_direction__01", "video_id": "MOSE__b826d642", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (brown duck with green beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "backward", "up", "down"], "frame_paths": ["media/9c30ea40f3679b4cce4d/F0.jpg", "media/9c30ea40f3679b4cce4d/F1.jpg", "media/9c30ea40f3679b4cce4d/F2.jpg", "media/9c30ea40f3679b4cce4d/F3.jpg", "media/9c30ea40f3679b4cce4d/F4.jpg", "media/9c30ea40f3679b4cce4d/F5.jpg", "media/9c30ea40f3679b4cce4d/F6.jpg", "media/9c30ea40f3679b4cce4d/F7.jpg", "media/9c30ea40f3679b4cce4d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (brown duck with green beak) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc64d27ed76b81e6f641416742ddad93cdd7e973dad7f3d89c7b45445708fc72"}
{"qa_id": "MOSE__b8651773__object_world_motion_direction__00", "video_id": "MOSE__b8651773", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (black bear standing near the rock wall on the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "right", "left"], "frame_paths": ["media/35f5bfd2221d2cf7c4a5/F0.jpg", "media/35f5bfd2221d2cf7c4a5/F1.jpg", "media/35f5bfd2221d2cf7c4a5/F2.jpg", "media/35f5bfd2221d2cf7c4a5/F3.jpg", "media/35f5bfd2221d2cf7c4a5/F4.jpg", "media/35f5bfd2221d2cf7c4a5/F5.jpg", "media/35f5bfd2221d2cf7c4a5/F6.jpg", "media/35f5bfd2221d2cf7c4a5/F7.jpg", "media/35f5bfd2221d2cf7c4a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (black bear standing near the rock wall on the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0cba65a00f9ff8841cedb83ca5097c3ec829be61312001f7906fe6cf5976b349"}
{"qa_id": "MOSE__b8651773__object_pair_distance_change__01", "video_id": "MOSE__b8651773", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did the 3D distance between object A (black bear standing near the rock wall on the left) and object B (black bear standing near the rock wall on the right) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4f636010a3689e623edf/F0.jpg", "media/4f636010a3689e623edf/F1.jpg", "media/4f636010a3689e623edf/F2.jpg", "media/4f636010a3689e623edf/F3.jpg", "media/4f636010a3689e623edf/F4.jpg", "media/4f636010a3689e623edf/F5.jpg", "media/4f636010a3689e623edf/F6.jpg", "media/4f636010a3689e623edf/F7.jpg", "media/4f636010a3689e623edf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (black bear standing near the rock wall on the left) and object B (black bear standing near the rock wall on the right) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efd221bd1d395d6dd6457ae80456e434832ca36563ab8c9cd81fbc318f374c47"}
{"qa_id": "MOSE__b8651773__object_travel_distance_comparison__02", "video_id": "MOSE__b8651773", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black bear standing near the rock wall on the left) or B (black bear standing near the rock wall on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4f636010a3689e623edf/F0.jpg", "media/4f636010a3689e623edf/F1.jpg", "media/4f636010a3689e623edf/F2.jpg", "media/4f636010a3689e623edf/F3.jpg", "media/4f636010a3689e623edf/F4.jpg", "media/4f636010a3689e623edf/F5.jpg", "media/4f636010a3689e623edf/F6.jpg", "media/4f636010a3689e623edf/F7.jpg", "media/4f636010a3689e623edf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black bear standing near the rock wall on the left) or B (black bear standing near the rock wall on the right)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3efe944c2cfb8b0f9e5b2db9d58db18d7d9a51620c1482d8f38d78033d61355f"}
{"qa_id": "MOSE__b942ce0a__object_world_motion_direction__00", "video_id": "MOSE__b942ce0a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (small white fish with orange fins) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/dd21989cefca4e368c4a/F0.jpg", "media/dd21989cefca4e368c4a/F1.jpg", "media/dd21989cefca4e368c4a/F2.jpg", "media/dd21989cefca4e368c4a/F3.jpg", "media/dd21989cefca4e368c4a/F4.jpg", "media/dd21989cefca4e368c4a/F5.jpg", "media/dd21989cefca4e368c4a/F6.jpg", "media/dd21989cefca4e368c4a/F7.jpg", "media/dd21989cefca4e368c4a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (small white fish with orange fins) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5e308d03a9add727de5074da37a71a6f709b9168165f593889c90edc8991320"}
{"qa_id": "MOSE__ba61a719__object_world_motion_direction__00", "video_id": "MOSE__ba61a719", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (motocross rider in orange jersey with number 8) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "left", "up"], "frame_paths": ["media/eb49d0cdbcd0d57e4cdb/F0.jpg", "media/eb49d0cdbcd0d57e4cdb/F1.jpg", "media/eb49d0cdbcd0d57e4cdb/F2.jpg", "media/eb49d0cdbcd0d57e4cdb/F3.jpg", "media/eb49d0cdbcd0d57e4cdb/F4.jpg", "media/eb49d0cdbcd0d57e4cdb/F5.jpg", "media/eb49d0cdbcd0d57e4cdb/F6.jpg", "media/eb49d0cdbcd0d57e4cdb/F7.jpg", "media/eb49d0cdbcd0d57e4cdb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (motocross rider in orange jersey with number 8) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5f19ee9dce4d0c85462b54d38d89b9b4b521f5b8b25e85b8196efc7f36bcef4"}
{"qa_id": "MOSE__bbe97d18__camera_translation_direction__00", "video_id": "MOSE__bbe97d18", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "up", "down", "right", "forward"], "frame_paths": ["media/0ccf00a5aa577469dec5/F0.jpg", "media/0ccf00a5aa577469dec5/F1.jpg", "media/0ccf00a5aa577469dec5/F2.jpg", "media/0ccf00a5aa577469dec5/F3.jpg", "media/0ccf00a5aa577469dec5/F4.jpg", "media/0ccf00a5aa577469dec5/F5.jpg", "media/0ccf00a5aa577469dec5/F6.jpg", "media/0ccf00a5aa577469dec5/F7.jpg", "media/0ccf00a5aa577469dec5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. up\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c7f40d5dd716c279904387f771d25cf826e358ec40ec8f7673b9c846bfad1574"}
{"qa_id": "MOSE__bcb6e338__object_world_motion_direction__00", "video_id": "MOSE__bcb6e338", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (long dark wooden boat with a roof) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/672e4e1bc66a1449dff2/F0.jpg", "media/672e4e1bc66a1449dff2/F1.jpg", "media/672e4e1bc66a1449dff2/F2.jpg", "media/672e4e1bc66a1449dff2/F3.jpg", "media/672e4e1bc66a1449dff2/F4.jpg", "media/672e4e1bc66a1449dff2/F5.jpg", "media/672e4e1bc66a1449dff2/F6.jpg", "media/672e4e1bc66a1449dff2/F7.jpg", "media/672e4e1bc66a1449dff2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (long dark wooden boat with a roof) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb8943140b3dce50a4d9c4e0be3bf1fa421a8bd8bbd3c8f02248aae487ca67ba"}
{"qa_id": "MOSE__beb693c9__camera_view_direction_change__00", "video_id": "MOSE__beb693c9", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/12b07a9ef8adb18e4a23/F0.jpg", "media/12b07a9ef8adb18e4a23/F1.jpg", "media/12b07a9ef8adb18e4a23/F2.jpg", "media/12b07a9ef8adb18e4a23/F3.jpg", "media/12b07a9ef8adb18e4a23/F4.jpg", "media/12b07a9ef8adb18e4a23/F5.jpg", "media/12b07a9ef8adb18e4a23/F6.jpg", "media/12b07a9ef8adb18e4a23/F7.jpg", "media/12b07a9ef8adb18e4a23/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "892b23147d17a5e53e289278ee769dfa3e247cd14585fe2cb2e4458cf11b33c7"}
{"qa_id": "MOSE__beb693c9__object_world_motion_direction__01", "video_id": "MOSE__beb693c9", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (large blue and white ferry with 'HELLENIC SEAFARER' text on the hull) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "down", "right", "left"], "frame_paths": ["media/d2adfd1c9b8f3178ad44/F0.jpg", "media/d2adfd1c9b8f3178ad44/F1.jpg", "media/d2adfd1c9b8f3178ad44/F2.jpg", "media/d2adfd1c9b8f3178ad44/F3.jpg", "media/d2adfd1c9b8f3178ad44/F4.jpg", "media/d2adfd1c9b8f3178ad44/F5.jpg", "media/d2adfd1c9b8f3178ad44/F6.jpg", "media/d2adfd1c9b8f3178ad44/F7.jpg", "media/d2adfd1c9b8f3178ad44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (large blue and white ferry with 'HELLENIC SEAFARER' text on the hull) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa2576bff7cf9160a3cf6fa64bd113e20e29b1b637a1573479ae02c6e5e3c2db"}
{"qa_id": "MOSE__c062f4d7__object_world_motion_direction__00", "video_id": "MOSE__c062f4d7", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (yellow fish with white patches) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/85407df55e100c96deda/F0.jpg", "media/85407df55e100c96deda/F1.jpg", "media/85407df55e100c96deda/F2.jpg", "media/85407df55e100c96deda/F3.jpg", "media/85407df55e100c96deda/F4.jpg", "media/85407df55e100c96deda/F5.jpg", "media/85407df55e100c96deda/F6.jpg", "media/85407df55e100c96deda/F7.jpg", "media/85407df55e100c96deda/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (yellow fish with white patches) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7bb081388dbad0bba7fc35a385dfd4886b044273a1a8064434ee727db4319363"}
{"qa_id": "MOSE__c09da3b2__camera_object_distance_change__00", "video_id": "MOSE__c09da3b2", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did object A (person in light grey t-shirt riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5ff699311ffa903e7384/F0.jpg", "media/5ff699311ffa903e7384/F1.jpg", "media/5ff699311ffa903e7384/F2.jpg", "media/5ff699311ffa903e7384/F3.jpg", "media/5ff699311ffa903e7384/F4.jpg", "media/5ff699311ffa903e7384/F5.jpg", "media/5ff699311ffa903e7384/F6.jpg", "media/5ff699311ffa903e7384/F7.jpg", "media/5ff699311ffa903e7384/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (person in light grey t-shirt riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4b0f82617a72b011fea59c6928f17a67dfdcc04f194bc0ce8ea9c281a0000afb"}
{"qa_id": "MOSE__c1ec4b26__camera_object_distance_change__00", "video_id": "MOSE__c1ec4b26", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object A (large grey elephant standing over a person) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/08401d5ec756732b13af/F0.jpg", "media/08401d5ec756732b13af/F1.jpg", "media/08401d5ec756732b13af/F2.jpg", "media/08401d5ec756732b13af/F3.jpg", "media/08401d5ec756732b13af/F4.jpg", "media/08401d5ec756732b13af/F5.jpg", "media/08401d5ec756732b13af/F6.jpg", "media/08401d5ec756732b13af/F7.jpg", "media/08401d5ec756732b13af/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (large grey elephant standing over a person) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a470090a5bb10fa866320024b014460a193776e3999913443fd21a1416e4cc7"}
{"qa_id": "MOSE__c28d81a9__object_world_motion_direction__00", "video_id": "MOSE__c28d81a9", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (grey tabby kitten) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/dfcfeeb9b7ffb642a9ff/F0.jpg", "media/dfcfeeb9b7ffb642a9ff/F1.jpg", "media/dfcfeeb9b7ffb642a9ff/F2.jpg", "media/dfcfeeb9b7ffb642a9ff/F3.jpg", "media/dfcfeeb9b7ffb642a9ff/F4.jpg", "media/dfcfeeb9b7ffb642a9ff/F5.jpg", "media/dfcfeeb9b7ffb642a9ff/F6.jpg", "media/dfcfeeb9b7ffb642a9ff/F7.jpg", "media/dfcfeeb9b7ffb642a9ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (grey tabby kitten) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e4fa18d4e9d93c985cf256d1d67a7b1593da18663d058cece7d926b35294e3a6"}
{"qa_id": "MOSE__c28d81a9__object_pair_distance_change__01", "video_id": "MOSE__c28d81a9", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F5, did the 3D distance between object A (grey tabby kitten) and object B (dark tabby cat tail) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/df71d72608c32d543400/F0.jpg", "media/df71d72608c32d543400/F1.jpg", "media/df71d72608c32d543400/F2.jpg", "media/df71d72608c32d543400/F3.jpg", "media/df71d72608c32d543400/F4.jpg", "media/df71d72608c32d543400/F5.jpg", "media/df71d72608c32d543400/F6.jpg", "media/df71d72608c32d543400/F7.jpg", "media/df71d72608c32d543400/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did the 3D distance between object A (grey tabby kitten) and object B (dark tabby cat tail) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c70c2d7ffa106aadc4b68218bb27b98d2b8b0bf5303758c395cece82e879e495"}
{"qa_id": "MOSE__c30641cf__object_pair_distance_change__00", "video_id": "MOSE__c30641cf", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (orange flamingo bending down to feed) and object B (white flamingo standing with neck curved) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d80be43dda42f1c2ace0/F0.jpg", "media/d80be43dda42f1c2ace0/F1.jpg", "media/d80be43dda42f1c2ace0/F2.jpg", "media/d80be43dda42f1c2ace0/F3.jpg", "media/d80be43dda42f1c2ace0/F4.jpg", "media/d80be43dda42f1c2ace0/F5.jpg", "media/d80be43dda42f1c2ace0/F6.jpg", "media/d80be43dda42f1c2ace0/F7.jpg", "media/d80be43dda42f1c2ace0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g3", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (orange flamingo bending down to feed) and object B (white flamingo standing with neck curved) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b04e346ead03be90a46bfddb5c9270bb49f79aa5abc95696ba0035a8b6b2c84c"}
{"qa_id": "MOSE__c33adbc2__camera_view_direction_change__00", "video_id": "MOSE__c33adbc2", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/dcfb160979e57db190a2/F0.jpg", "media/dcfb160979e57db190a2/F1.jpg", "media/dcfb160979e57db190a2/F2.jpg", "media/dcfb160979e57db190a2/F3.jpg", "media/dcfb160979e57db190a2/F4.jpg", "media/dcfb160979e57db190a2/F5.jpg", "media/dcfb160979e57db190a2/F6.jpg", "media/dcfb160979e57db190a2/F7.jpg", "media/dcfb160979e57db190a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ba958ba5699228be3fb42e7f18a5cb11bdc27379be83cdbeb1d2e55c0e680044"}
{"qa_id": "MOSE__c4957274__camera_object_distance_change__00", "video_id": "MOSE__c4957274", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (person in dark coat walking on the street) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2a66ae8cd1feb3d7a3df/F0.jpg", "media/2a66ae8cd1feb3d7a3df/F1.jpg", "media/2a66ae8cd1feb3d7a3df/F2.jpg", "media/2a66ae8cd1feb3d7a3df/F3.jpg", "media/2a66ae8cd1feb3d7a3df/F4.jpg", "media/2a66ae8cd1feb3d7a3df/F5.jpg", "media/2a66ae8cd1feb3d7a3df/F6.jpg", "media/2a66ae8cd1feb3d7a3df/F7.jpg", "media/2a66ae8cd1feb3d7a3df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (person in dark coat walking on the street) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebd475566d39443c43c87f0ce3e68c3566d3e1dc3d0cccab1167281dfc489f86"}
{"qa_id": "MOSE__c4957274__camera_object_distance_change__01", "video_id": "MOSE__c4957274", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object B (white suv driving on the road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c5ba1d0ee7827cc865ad/F0.jpg", "media/c5ba1d0ee7827cc865ad/F1.jpg", "media/c5ba1d0ee7827cc865ad/F2.jpg", "media/c5ba1d0ee7827cc865ad/F3.jpg", "media/c5ba1d0ee7827cc865ad/F4.jpg", "media/c5ba1d0ee7827cc865ad/F5.jpg", "media/c5ba1d0ee7827cc865ad/F6.jpg", "media/c5ba1d0ee7827cc865ad/F7.jpg", "media/c5ba1d0ee7827cc865ad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object B (white suv driving on the road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e4f5738d0207553d9af3358df804df70e2c6f81722703e66e21ccf1115bd8ae7"}
{"qa_id": "MOSE__c4a0ecfb__object_world_motion_direction__00", "video_id": "MOSE__c4a0ecfb", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (white passenger van) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "forward", "backward", "up"], "frame_paths": ["media/975e28d0207ad5eefb5a/F0.jpg", "media/975e28d0207ad5eefb5a/F1.jpg", "media/975e28d0207ad5eefb5a/F2.jpg", "media/975e28d0207ad5eefb5a/F3.jpg", "media/975e28d0207ad5eefb5a/F4.jpg", "media/975e28d0207ad5eefb5a/F5.jpg", "media/975e28d0207ad5eefb5a/F6.jpg", "media/975e28d0207ad5eefb5a/F7.jpg", "media/975e28d0207ad5eefb5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (white passenger van) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5bddea8379862fef6e3f332c1eba368886c1b81dad260bc14b436106abb8ce9e"}
{"qa_id": "MOSE__c4a0ecfb__camera_object_distance_change__01", "video_id": "MOSE__c4a0ecfb", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (white passenger van) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/975e28d0207ad5eefb5a/F0.jpg", "media/975e28d0207ad5eefb5a/F1.jpg", "media/975e28d0207ad5eefb5a/F2.jpg", "media/975e28d0207ad5eefb5a/F3.jpg", "media/975e28d0207ad5eefb5a/F4.jpg", "media/975e28d0207ad5eefb5a/F5.jpg", "media/975e28d0207ad5eefb5a/F6.jpg", "media/975e28d0207ad5eefb5a/F7.jpg", "media/975e28d0207ad5eefb5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (white passenger van) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f93df86dd2c7b3c8b4e51154f8d073b6d6b26176eee1fa91ca3169944b98fd36"}
{"qa_id": "MOSE__c4aa176c__camera_translation_direction__00", "video_id": "MOSE__c4aa176c", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/ddab482cffb8fefb93c8/F0.jpg", "media/ddab482cffb8fefb93c8/F1.jpg", "media/ddab482cffb8fefb93c8/F2.jpg", "media/ddab482cffb8fefb93c8/F3.jpg", "media/ddab482cffb8fefb93c8/F4.jpg", "media/ddab482cffb8fefb93c8/F5.jpg", "media/ddab482cffb8fefb93c8/F6.jpg", "media/ddab482cffb8fefb93c8/F7.jpg", "media/ddab482cffb8fefb93c8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5cd3f8706a92139f6f8505332a5cc4a4f13b67492be677dc9e6aadb17de7d2d"}
{"qa_id": "MOSE__c4aa176c__object_travel_distance_comparison__02", "video_id": "MOSE__c4aa176c", "task": "object_travel_distance_comparison", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black tabby cat) or B (orange and white cat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f9f651390f3cb147ba44/F0.jpg", "media/f9f651390f3cb147ba44/F1.jpg", "media/f9f651390f3cb147ba44/F2.jpg", "media/f9f651390f3cb147ba44/F3.jpg", "media/f9f651390f3cb147ba44/F4.jpg", "media/f9f651390f3cb147ba44/F5.jpg", "media/f9f651390f3cb147ba44/F6.jpg", "media/f9f651390f3cb147ba44/F7.jpg", "media/f9f651390f3cb147ba44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black tabby cat) or B (orange and white cat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2ebcc990f8f04b27f5f1efbe88383c7f5a3551fb13a0daac3ea5a9a11bc164a6"}
{"qa_id": "MOSE__c4f4b2e2__object_world_motion_direction__00", "video_id": "MOSE__c4f4b2e2", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (grey and white goose with a long neck) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "forward", "backward"], "frame_paths": ["media/9b4c9c14aeaca6b4cf19/F0.jpg", "media/9b4c9c14aeaca6b4cf19/F1.jpg", "media/9b4c9c14aeaca6b4cf19/F2.jpg", "media/9b4c9c14aeaca6b4cf19/F3.jpg", "media/9b4c9c14aeaca6b4cf19/F4.jpg", "media/9b4c9c14aeaca6b4cf19/F5.jpg", "media/9b4c9c14aeaca6b4cf19/F6.jpg", "media/9b4c9c14aeaca6b4cf19/F7.jpg", "media/9b4c9c14aeaca6b4cf19/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g7"}, "requested_object_ids": ["g7"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (grey and white goose with a long neck) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53d988da20075728c69928d919ca69aea1e95d12432a751e06b1aa9a443f911"}
{"qa_id": "MOSE__c6726eb8__object_world_motion_direction__00", "video_id": "MOSE__c6726eb8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (black compact SUV) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/81b80fc15de8289fd659/F0.jpg", "media/81b80fc15de8289fd659/F1.jpg", "media/81b80fc15de8289fd659/F2.jpg", "media/81b80fc15de8289fd659/F3.jpg", "media/81b80fc15de8289fd659/F4.jpg", "media/81b80fc15de8289fd659/F5.jpg", "media/81b80fc15de8289fd659/F6.jpg", "media/81b80fc15de8289fd659/F7.jpg", "media/81b80fc15de8289fd659/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (black compact SUV) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2f386ffed61ee941c8952903f615742963a231957fe501c40beb9c822c6b1587"}
{"qa_id": "MOSE__c6726eb8__camera_object_distance_change__01", "video_id": "MOSE__c6726eb8", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object A (black compact SUV) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/81b80fc15de8289fd659/F0.jpg", "media/81b80fc15de8289fd659/F1.jpg", "media/81b80fc15de8289fd659/F2.jpg", "media/81b80fc15de8289fd659/F3.jpg", "media/81b80fc15de8289fd659/F4.jpg", "media/81b80fc15de8289fd659/F5.jpg", "media/81b80fc15de8289fd659/F6.jpg", "media/81b80fc15de8289fd659/F7.jpg", "media/81b80fc15de8289fd659/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (black compact SUV) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a959455d0864c339e92970b1a64fb72685b42b6176903714ad58b0faac81867"}
{"qa_id": "MOSE__c6726eb8__object_world_motion_direction__02", "video_id": "MOSE__c6726eb8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (person riding a scooter) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "right", "left", "down"], "frame_paths": ["media/47f2ba3656c13ff2097a/F0.jpg", "media/47f2ba3656c13ff2097a/F1.jpg", "media/47f2ba3656c13ff2097a/F2.jpg", "media/47f2ba3656c13ff2097a/F3.jpg", "media/47f2ba3656c13ff2097a/F4.jpg", "media/47f2ba3656c13ff2097a/F5.jpg", "media/47f2ba3656c13ff2097a/F6.jpg", "media/47f2ba3656c13ff2097a/F7.jpg", "media/47f2ba3656c13ff2097a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (person riding a scooter) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1956d839fb562d3cc291447d6e0b5d831be991502825156c2608a232b38fa0a4"}
{"qa_id": "MOSE__c6726eb8__camera_object_distance_change__03", "video_id": "MOSE__c6726eb8", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object B (person riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/47f2ba3656c13ff2097a/F0.jpg", "media/47f2ba3656c13ff2097a/F1.jpg", "media/47f2ba3656c13ff2097a/F2.jpg", "media/47f2ba3656c13ff2097a/F3.jpg", "media/47f2ba3656c13ff2097a/F4.jpg", "media/47f2ba3656c13ff2097a/F5.jpg", "media/47f2ba3656c13ff2097a/F6.jpg", "media/47f2ba3656c13ff2097a/F7.jpg", "media/47f2ba3656c13ff2097a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object B (person riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e614f73ce681c57f1e522797b9f16667415bda94e556e168dce83eb0e08ca1be"}
{"qa_id": "MOSE__c6726eb8__object_pair_distance_change__04", "video_id": "MOSE__c6726eb8", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (black compact SUV) and object B (person riding a scooter) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d8884d3ebabe3d8d5888/F0.jpg", "media/d8884d3ebabe3d8d5888/F1.jpg", "media/d8884d3ebabe3d8d5888/F2.jpg", "media/d8884d3ebabe3d8d5888/F3.jpg", "media/d8884d3ebabe3d8d5888/F4.jpg", "media/d8884d3ebabe3d8d5888/F5.jpg", "media/d8884d3ebabe3d8d5888/F6.jpg", "media/d8884d3ebabe3d8d5888/F7.jpg", "media/d8884d3ebabe3d8d5888/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (black compact SUV) and object B (person riding a scooter) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "18bd1f8967bc653759c2e723aa1e1caca947cfde41c962982ca36b8203b4ea21"}
{"qa_id": "MOSE__c6726eb8__object_travel_distance_comparison__05", "video_id": "MOSE__c6726eb8", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black compact SUV) or B (person riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/d8884d3ebabe3d8d5888/F0.jpg", "media/d8884d3ebabe3d8d5888/F1.jpg", "media/d8884d3ebabe3d8d5888/F2.jpg", "media/d8884d3ebabe3d8d5888/F3.jpg", "media/d8884d3ebabe3d8d5888/F4.jpg", "media/d8884d3ebabe3d8d5888/F5.jpg", "media/d8884d3ebabe3d8d5888/F6.jpg", "media/d8884d3ebabe3d8d5888/F7.jpg", "media/d8884d3ebabe3d8d5888/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black compact SUV) or B (person riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f40605556e612a7d66b994833e814e75a500b21bb0f4ddd78d332ddf0a74bfd9"}
{"qa_id": "MOSE__c7e745ed__object_world_motion_direction__00", "video_id": "MOSE__c7e745ed", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (brown cow with 'A009' written on its side) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "down", "right", "left"], "frame_paths": ["media/cf70a4d1831b8a413a3e/F0.jpg", "media/cf70a4d1831b8a413a3e/F1.jpg", "media/cf70a4d1831b8a413a3e/F2.jpg", "media/cf70a4d1831b8a413a3e/F3.jpg", "media/cf70a4d1831b8a413a3e/F4.jpg", "media/cf70a4d1831b8a413a3e/F5.jpg", "media/cf70a4d1831b8a413a3e/F6.jpg", "media/cf70a4d1831b8a413a3e/F7.jpg", "media/cf70a4d1831b8a413a3e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (brown cow with 'A009' written on its side) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c4dce697ac0e9902911ee781a2f8486b7ecaa2751b82463a9d7331af64163be3"}
{"qa_id": "MOSE__c7fe0584__object_travel_distance_comparison__00", "video_id": "MOSE__c7fe0584", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in maroon traditional robe with sunglasses hanging from collar) or B (man in grey plaid suit holding a bouquet of red roses)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/6c84151c8201772c7331/F0.jpg", "media/6c84151c8201772c7331/F1.jpg", "media/6c84151c8201772c7331/F2.jpg", "media/6c84151c8201772c7331/F3.jpg", "media/6c84151c8201772c7331/F4.jpg", "media/6c84151c8201772c7331/F5.jpg", "media/6c84151c8201772c7331/F6.jpg", "media/6c84151c8201772c7331/F7.jpg", "media/6c84151c8201772c7331/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in maroon traditional robe with sunglasses hanging from collar) or B (man in grey plaid suit holding a bouquet of red roses)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff35125b970743d4ed6dac6754e12e6d982a79e08151cee0c305337ad625bd70"}
{"qa_id": "MOSE__c8f0fd41__camera_translation_direction__00", "video_id": "MOSE__c8f0fd41", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/d711dfcb2bcd8f43a986/F0.jpg", "media/d711dfcb2bcd8f43a986/F1.jpg", "media/d711dfcb2bcd8f43a986/F2.jpg", "media/d711dfcb2bcd8f43a986/F3.jpg", "media/d711dfcb2bcd8f43a986/F4.jpg", "media/d711dfcb2bcd8f43a986/F5.jpg", "media/d711dfcb2bcd8f43a986/F6.jpg", "media/d711dfcb2bcd8f43a986/F7.jpg", "media/d711dfcb2bcd8f43a986/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c43b2f70fa65cab8d84396899ab4af7adbecd61cd592a105e418c6e3a5b91171"}
{"qa_id": "MOSE__c9351029__camera_view_direction_change__00", "video_id": "MOSE__c9351029", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/b9a5a4875831964685d2/F0.jpg", "media/b9a5a4875831964685d2/F1.jpg", "media/b9a5a4875831964685d2/F2.jpg", "media/b9a5a4875831964685d2/F3.jpg", "media/b9a5a4875831964685d2/F4.jpg", "media/b9a5a4875831964685d2/F5.jpg", "media/b9a5a4875831964685d2/F6.jpg", "media/b9a5a4875831964685d2/F7.jpg", "media/b9a5a4875831964685d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ace13b32caef12f3fec7578f549de518b59622a94b4a701755c331166994d961"}
{"qa_id": "MOSE__c9351029__object_world_motion_direction__02", "video_id": "MOSE__c9351029", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (brown cow with a white face walking on the path) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "right", "up", "backward"], "frame_paths": ["media/462ec3cf0769e238ad1d/F0.jpg", "media/462ec3cf0769e238ad1d/F1.jpg", "media/462ec3cf0769e238ad1d/F2.jpg", "media/462ec3cf0769e238ad1d/F3.jpg", "media/462ec3cf0769e238ad1d/F4.jpg", "media/462ec3cf0769e238ad1d/F5.jpg", "media/462ec3cf0769e238ad1d/F6.jpg", "media/462ec3cf0769e238ad1d/F7.jpg", "media/462ec3cf0769e238ad1d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (brown cow with a white face walking on the path) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "328a587f68b78a6a7dcfe98d1f40ee8be0cfe10f77ca7341cb103e1858502880"}
{"qa_id": "MOSE__c9351029__camera_object_distance_change__03", "video_id": "MOSE__c9351029", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (brown cow with a white face walking on the path) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/462ec3cf0769e238ad1d/F0.jpg", "media/462ec3cf0769e238ad1d/F1.jpg", "media/462ec3cf0769e238ad1d/F2.jpg", "media/462ec3cf0769e238ad1d/F3.jpg", "media/462ec3cf0769e238ad1d/F4.jpg", "media/462ec3cf0769e238ad1d/F5.jpg", "media/462ec3cf0769e238ad1d/F6.jpg", "media/462ec3cf0769e238ad1d/F7.jpg", "media/462ec3cf0769e238ad1d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (brown cow with a white face walking on the path) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f317658bb5ede326c057314fe8d4239b8da2330edc2db539cbca6d771b96903"}
{"qa_id": "MOSE__c9f1d87f__object_travel_distance_comparison__00", "video_id": "MOSE__c9f1d87f", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey plaid suit) or B (man in maroon traditional outfit)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/9932e86a2a7e7f475973/F0.jpg", "media/9932e86a2a7e7f475973/F1.jpg", "media/9932e86a2a7e7f475973/F2.jpg", "media/9932e86a2a7e7f475973/F3.jpg", "media/9932e86a2a7e7f475973/F4.jpg", "media/9932e86a2a7e7f475973/F5.jpg", "media/9932e86a2a7e7f475973/F6.jpg", "media/9932e86a2a7e7f475973/F7.jpg", "media/9932e86a2a7e7f475973/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey plaid suit) or B (man in maroon traditional outfit)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eb0e020c48c37ff2be554035888c2234aca5d977fe78c133fa00f7d414e27a42"}
{"qa_id": "MOSE__ca619b7f__object_world_motion_direction__00", "video_id": "MOSE__ca619b7f", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person wearing a long red coat) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "down", "forward", "up"], "frame_paths": ["media/1d9e29ce0ab66540e23e/F0.jpg", "media/1d9e29ce0ab66540e23e/F1.jpg", "media/1d9e29ce0ab66540e23e/F2.jpg", "media/1d9e29ce0ab66540e23e/F3.jpg", "media/1d9e29ce0ab66540e23e/F4.jpg", "media/1d9e29ce0ab66540e23e/F5.jpg", "media/1d9e29ce0ab66540e23e/F6.jpg", "media/1d9e29ce0ab66540e23e/F7.jpg", "media/1d9e29ce0ab66540e23e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person wearing a long red coat) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. down\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b03981dc923f157c86b24a78aae498b5490d4f83c9f8596339ed6cb00a38a12"}
{"qa_id": "MOSE__cb170015__camera_view_direction_change__00", "video_id": "MOSE__cb170015", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/2fb92217cfc377ba5ec9/F0.jpg", "media/2fb92217cfc377ba5ec9/F1.jpg", "media/2fb92217cfc377ba5ec9/F2.jpg", "media/2fb92217cfc377ba5ec9/F3.jpg", "media/2fb92217cfc377ba5ec9/F4.jpg", "media/2fb92217cfc377ba5ec9/F5.jpg", "media/2fb92217cfc377ba5ec9/F6.jpg", "media/2fb92217cfc377ba5ec9/F7.jpg", "media/2fb92217cfc377ba5ec9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4c234c17234347490ac73669d876f5dc4e1e7ffdabfeadf8f0d17fcba1dfb923"}
{"qa_id": "MOSE__cb8eedf4__camera_view_direction_change__00", "video_id": "MOSE__cb8eedf4", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/b8e6d446e5831dd27209/F0.jpg", "media/b8e6d446e5831dd27209/F1.jpg", "media/b8e6d446e5831dd27209/F2.jpg", "media/b8e6d446e5831dd27209/F3.jpg", "media/b8e6d446e5831dd27209/F4.jpg", "media/b8e6d446e5831dd27209/F5.jpg", "media/b8e6d446e5831dd27209/F6.jpg", "media/b8e6d446e5831dd27209/F7.jpg", "media/b8e6d446e5831dd27209/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ba958ba5699228be3fb42e7f18a5cb11bdc27379be83cdbeb1d2e55c0e680044"}
{"qa_id": "MOSE__cb8eedf4__object_world_motion_direction__01", "video_id": "MOSE__cb8eedf4", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (brown horse with a dark mane) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "forward", "backward"], "frame_paths": ["media/5a1e76b20e2f73c8f42a/F0.jpg", "media/5a1e76b20e2f73c8f42a/F1.jpg", "media/5a1e76b20e2f73c8f42a/F2.jpg", "media/5a1e76b20e2f73c8f42a/F3.jpg", "media/5a1e76b20e2f73c8f42a/F4.jpg", "media/5a1e76b20e2f73c8f42a/F5.jpg", "media/5a1e76b20e2f73c8f42a/F6.jpg", "media/5a1e76b20e2f73c8f42a/F7.jpg", "media/5a1e76b20e2f73c8f42a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (brown horse with a dark mane) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0aee67c281d4c3107b70f7fc42374a4435dff0ef292d923194fc6ad9926aa2c7"}
{"qa_id": "MOSE__cbd144f5__object_world_motion_direction__00", "video_id": "MOSE__cbd144f5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (large giraffe with brown and cream patterned coat walking towards the camera) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "right", "forward", "backward"], "frame_paths": ["media/11700483c580f2339eca/F0.jpg", "media/11700483c580f2339eca/F1.jpg", "media/11700483c580f2339eca/F2.jpg", "media/11700483c580f2339eca/F3.jpg", "media/11700483c580f2339eca/F4.jpg", "media/11700483c580f2339eca/F5.jpg", "media/11700483c580f2339eca/F6.jpg", "media/11700483c580f2339eca/F7.jpg", "media/11700483c580f2339eca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (large giraffe with brown and cream patterned coat walking towards the camera) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8994da4c23d45b900c14015b4bd7ff8e5d409b25d4df8d6241622076866c7022"}
{"qa_id": "MOSE__cbd144f5__camera_object_distance_change__01", "video_id": "MOSE__cbd144f5", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (large giraffe with brown and cream patterned coat walking towards the camera) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/11700483c580f2339eca/F0.jpg", "media/11700483c580f2339eca/F1.jpg", "media/11700483c580f2339eca/F2.jpg", "media/11700483c580f2339eca/F3.jpg", "media/11700483c580f2339eca/F4.jpg", "media/11700483c580f2339eca/F5.jpg", "media/11700483c580f2339eca/F6.jpg", "media/11700483c580f2339eca/F7.jpg", "media/11700483c580f2339eca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (large giraffe with brown and cream patterned coat walking towards the camera) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5945be24f0240dcb8221172981a0228c350789ad336b5e7ccd091b262cd92cdd"}
{"qa_id": "MOSE__ce9a2796__object_world_motion_direction__00", "video_id": "MOSE__ce9a2796", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (large grey elephant) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "down", "left", "forward"], "frame_paths": ["media/e50fb8ccd4ec8880ab6c/F0.jpg", "media/e50fb8ccd4ec8880ab6c/F1.jpg", "media/e50fb8ccd4ec8880ab6c/F2.jpg", "media/e50fb8ccd4ec8880ab6c/F3.jpg", "media/e50fb8ccd4ec8880ab6c/F4.jpg", "media/e50fb8ccd4ec8880ab6c/F5.jpg", "media/e50fb8ccd4ec8880ab6c/F6.jpg", "media/e50fb8ccd4ec8880ab6c/F7.jpg", "media/e50fb8ccd4ec8880ab6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (large grey elephant) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebaa3e9c0a6b1af80e36f1b444e7d531bbbf7bd3bb568e52ff44c5f1ee665b03"}
{"qa_id": "MOSE__ce9a2796__object_pair_distance_change__01", "video_id": "MOSE__ce9a2796", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (large grey elephant) and object B (man in dark blue t-shirt sitting on grass) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f5a9ca35c3fcb83d1689/F0.jpg", "media/f5a9ca35c3fcb83d1689/F1.jpg", "media/f5a9ca35c3fcb83d1689/F2.jpg", "media/f5a9ca35c3fcb83d1689/F3.jpg", "media/f5a9ca35c3fcb83d1689/F4.jpg", "media/f5a9ca35c3fcb83d1689/F5.jpg", "media/f5a9ca35c3fcb83d1689/F6.jpg", "media/f5a9ca35c3fcb83d1689/F7.jpg", "media/f5a9ca35c3fcb83d1689/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (large grey elephant) and object B (man in dark blue t-shirt sitting on grass) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73dc1182ef023782c4f1c2e99057330743ece07d18843628304cec33c8d85297"}
{"qa_id": "MOSE__ce9a2796__object_travel_distance_comparison__02", "video_id": "MOSE__ce9a2796", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant) or B (man in dark blue t-shirt sitting on grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f5a9ca35c3fcb83d1689/F0.jpg", "media/f5a9ca35c3fcb83d1689/F1.jpg", "media/f5a9ca35c3fcb83d1689/F2.jpg", "media/f5a9ca35c3fcb83d1689/F3.jpg", "media/f5a9ca35c3fcb83d1689/F4.jpg", "media/f5a9ca35c3fcb83d1689/F5.jpg", "media/f5a9ca35c3fcb83d1689/F6.jpg", "media/f5a9ca35c3fcb83d1689/F7.jpg", "media/f5a9ca35c3fcb83d1689/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant) or B (man in dark blue t-shirt sitting on grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb2e96405cdd95cb7c476fe095e9f99d90c20bdbe5c0118aaf4fd5d26ba92ab3"}
{"qa_id": "MOSE__ceaa6c65__camera_object_distance_change__00", "video_id": "MOSE__ceaa6c65", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did object A (silver sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/54c6040af546e05a77f1/F0.jpg", "media/54c6040af546e05a77f1/F1.jpg", "media/54c6040af546e05a77f1/F2.jpg", "media/54c6040af546e05a77f1/F3.jpg", "media/54c6040af546e05a77f1/F4.jpg", "media/54c6040af546e05a77f1/F5.jpg", "media/54c6040af546e05a77f1/F6.jpg", "media/54c6040af546e05a77f1/F7.jpg", "media/54c6040af546e05a77f1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (silver sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7917848e9a24fa16b8db745c58d5cdbcd8c846ee66c21cd4ef1606e147b6070"}
{"qa_id": "MOSE__cf376cf1__object_travel_distance_comparison__00", "video_id": "MOSE__cf376cf1", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with a light brown coat crossing the road on the right) or B (sheep with a light brown coat crossing the road on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/72833e9d5ef3d39f1c65/F0.jpg", "media/72833e9d5ef3d39f1c65/F1.jpg", "media/72833e9d5ef3d39f1c65/F2.jpg", "media/72833e9d5ef3d39f1c65/F3.jpg", "media/72833e9d5ef3d39f1c65/F4.jpg", "media/72833e9d5ef3d39f1c65/F5.jpg", "media/72833e9d5ef3d39f1c65/F6.jpg", "media/72833e9d5ef3d39f1c65/F7.jpg", "media/72833e9d5ef3d39f1c65/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (sheep with a light brown coat crossing the road on the right) or B (sheep with a light brown coat crossing the road on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab0824c5c881e88f950f6c8a7ad63843062f228ba436ccbd82b8514fcfe15639"}
{"qa_id": "MOSE__cfc68a82__camera_translation_direction__00", "video_id": "MOSE__cfc68a82", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "forward", "left", "right"], "frame_paths": ["media/3144c3ffb23a52d956c6/F0.jpg", "media/3144c3ffb23a52d956c6/F1.jpg", "media/3144c3ffb23a52d956c6/F2.jpg", "media/3144c3ffb23a52d956c6/F3.jpg", "media/3144c3ffb23a52d956c6/F4.jpg", "media/3144c3ffb23a52d956c6/F5.jpg", "media/3144c3ffb23a52d956c6/F6.jpg", "media/3144c3ffb23a52d956c6/F7.jpg", "media/3144c3ffb23a52d956c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "980a46948101b680c1a764696ced0c65062e3763833345f9c97cbcbe81f79517"}
{"qa_id": "MOSE__cfc68a82__camera_object_distance_change__01", "video_id": "MOSE__cfc68a82", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (monkey with brown fur and long tail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/378583034c775a3ed9b7/F0.jpg", "media/378583034c775a3ed9b7/F1.jpg", "media/378583034c775a3ed9b7/F2.jpg", "media/378583034c775a3ed9b7/F3.jpg", "media/378583034c775a3ed9b7/F4.jpg", "media/378583034c775a3ed9b7/F5.jpg", "media/378583034c775a3ed9b7/F6.jpg", "media/378583034c775a3ed9b7/F7.jpg", "media/378583034c775a3ed9b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (monkey with brown fur and long tail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab0a4df967ec050c55e448b87a1b3267123af4b70bf2f87a2e157ffb4650c22f"}
{"qa_id": "MOSE__cfc68a82__camera_object_distance_change__02", "video_id": "MOSE__cfc68a82", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object B (monkey with light brown fur sitting on the right) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/eb0cb8fd61210fa298ac/F0.jpg", "media/eb0cb8fd61210fa298ac/F1.jpg", "media/eb0cb8fd61210fa298ac/F2.jpg", "media/eb0cb8fd61210fa298ac/F3.jpg", "media/eb0cb8fd61210fa298ac/F4.jpg", "media/eb0cb8fd61210fa298ac/F5.jpg", "media/eb0cb8fd61210fa298ac/F6.jpg", "media/eb0cb8fd61210fa298ac/F7.jpg", "media/eb0cb8fd61210fa298ac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object B (monkey with light brown fur sitting on the right) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9d022e7f87a8ba42f0732f39e04469b25e0141bd017d86e7e85264ebcabcbf3b"}
{"qa_id": "MOSE__cfd946b2__object_pair_distance_change__00", "video_id": "MOSE__cfd946b2", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did the 3D distance between object A (koi fish with white body and black markings) and object B (orange koi fish) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8ad97348e13faf2ba3b5/F0.jpg", "media/8ad97348e13faf2ba3b5/F1.jpg", "media/8ad97348e13faf2ba3b5/F2.jpg", "media/8ad97348e13faf2ba3b5/F3.jpg", "media/8ad97348e13faf2ba3b5/F4.jpg", "media/8ad97348e13faf2ba3b5/F5.jpg", "media/8ad97348e13faf2ba3b5/F6.jpg", "media/8ad97348e13faf2ba3b5/F7.jpg", "media/8ad97348e13faf2ba3b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (koi fish with white body and black markings) and object B (orange koi fish) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c1edf15db59fbd753048731dad6b0a4f2c8f5575cdfeb766df7f25a1e9781b0"}
{"qa_id": "MOSE__cfd946b2__object_travel_distance_comparison__01", "video_id": "MOSE__cfd946b2", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (koi fish with white body and black markings) or B (orange koi fish)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/8ad97348e13faf2ba3b5/F0.jpg", "media/8ad97348e13faf2ba3b5/F1.jpg", "media/8ad97348e13faf2ba3b5/F2.jpg", "media/8ad97348e13faf2ba3b5/F3.jpg", "media/8ad97348e13faf2ba3b5/F4.jpg", "media/8ad97348e13faf2ba3b5/F5.jpg", "media/8ad97348e13faf2ba3b5/F6.jpg", "media/8ad97348e13faf2ba3b5/F7.jpg", "media/8ad97348e13faf2ba3b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (koi fish with white body and black markings) or B (orange koi fish)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cdb3423f46581fb1b13bf359e5904be80b3d567e754f628beb8b0ce813661126"}
{"qa_id": "MOSE__d07ad1be__object_world_motion_direction__00", "video_id": "MOSE__d07ad1be", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object A (large adult elephant walking on the paved road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "forward", "down", "left"], "frame_paths": ["media/516391fc1b654992089e/F0.jpg", "media/516391fc1b654992089e/F1.jpg", "media/516391fc1b654992089e/F2.jpg", "media/516391fc1b654992089e/F3.jpg", "media/516391fc1b654992089e/F4.jpg", "media/516391fc1b654992089e/F5.jpg", "media/516391fc1b654992089e/F6.jpg", "media/516391fc1b654992089e/F7.jpg", "media/516391fc1b654992089e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object A (large adult elephant walking on the paved road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d4a9f7acacdb00ed05e2e2e13a1b5402cc8c024c820ea6763d88809028be4b63"}
{"qa_id": "MOSE__d07ad1be__object_world_motion_direction__01", "video_id": "MOSE__d07ad1be", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object B (large adult elephant walking on the dirt shoulder to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "down", "right", "up", "left"], "frame_paths": ["media/d44c60ab7cac047b4e1b/F0.jpg", "media/d44c60ab7cac047b4e1b/F1.jpg", "media/d44c60ab7cac047b4e1b/F2.jpg", "media/d44c60ab7cac047b4e1b/F3.jpg", "media/d44c60ab7cac047b4e1b/F4.jpg", "media/d44c60ab7cac047b4e1b/F5.jpg", "media/d44c60ab7cac047b4e1b/F6.jpg", "media/d44c60ab7cac047b4e1b/F7.jpg", "media/d44c60ab7cac047b4e1b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object B (large adult elephant walking on the dirt shoulder to the right) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. down\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8b7379dd8d458a7eda7339c2d277f3d3b804599dc4df623017626732a645b45"}
{"qa_id": "MOSE__d1d95c61__object_world_motion_direction__00", "video_id": "MOSE__d1d95c61", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/94a503064ff169d287a7/F0.jpg", "media/94a503064ff169d287a7/F1.jpg", "media/94a503064ff169d287a7/F2.jpg", "media/94a503064ff169d287a7/F3.jpg", "media/94a503064ff169d287a7/F4.jpg", "media/94a503064ff169d287a7/F5.jpg", "media/94a503064ff169d287a7/F6.jpg", "media/94a503064ff169d287a7/F7.jpg", "media/94a503064ff169d287a7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (large adult elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6fcccd7b2f14409add23c9961aeead42c3bca4e8816bebc57cec86f5aeb83cd2"}
{"qa_id": "MOSE__d1d95c61__object_world_motion_direction__01", "video_id": "MOSE__d1d95c61", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object B (small juvenile elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "down", "up", "forward"], "frame_paths": ["media/3331180011f099f849d1/F0.jpg", "media/3331180011f099f849d1/F1.jpg", "media/3331180011f099f849d1/F2.jpg", "media/3331180011f099f849d1/F3.jpg", "media/3331180011f099f849d1/F4.jpg", "media/3331180011f099f849d1/F5.jpg", "media/3331180011f099f849d1/F6.jpg", "media/3331180011f099f849d1/F7.jpg", "media/3331180011f099f849d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object B (small juvenile elephant walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2194ca64ed3587ac09f03020b6ca9dd6d6568cd5fa5d1a2fce50f0340442b932"}
{"qa_id": "MOSE__d25ff89d__camera_object_distance_change__00", "video_id": "MOSE__d25ff89d", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object B (giant panda on the left side of the frame) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fac0b6785eda9b184551/F0.jpg", "media/fac0b6785eda9b184551/F1.jpg", "media/fac0b6785eda9b184551/F2.jpg", "media/fac0b6785eda9b184551/F3.jpg", "media/fac0b6785eda9b184551/F4.jpg", "media/fac0b6785eda9b184551/F5.jpg", "media/fac0b6785eda9b184551/F6.jpg", "media/fac0b6785eda9b184551/F7.jpg", "media/fac0b6785eda9b184551/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (giant panda on the left side of the frame) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "933b395d5bc0511deede2b6bfd9b25d0d36d6a743f3f07379d0ed370e916d30c"}
{"qa_id": "MOSE__d25ff89d__object_travel_distance_comparison__01", "video_id": "MOSE__d25ff89d", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (giant panda on the right side of the frame) or B (giant panda on the left side of the frame)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/2393623e1ea916c1fce9/F0.jpg", "media/2393623e1ea916c1fce9/F1.jpg", "media/2393623e1ea916c1fce9/F2.jpg", "media/2393623e1ea916c1fce9/F3.jpg", "media/2393623e1ea916c1fce9/F4.jpg", "media/2393623e1ea916c1fce9/F5.jpg", "media/2393623e1ea916c1fce9/F6.jpg", "media/2393623e1ea916c1fce9/F7.jpg", "media/2393623e1ea916c1fce9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (giant panda on the right side of the frame) or B (giant panda on the left side of the frame)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d01c296751f547d5b135bcaef46be36705fa6fd8fe845fb7a34e89f12986836"}
{"qa_id": "MOSE__d262b7e4__camera_object_distance_change__00", "video_id": "MOSE__d262b7e4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object A (white minivan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a5b44ff831d659e3772f/F0.jpg", "media/a5b44ff831d659e3772f/F1.jpg", "media/a5b44ff831d659e3772f/F2.jpg", "media/a5b44ff831d659e3772f/F3.jpg", "media/a5b44ff831d659e3772f/F4.jpg", "media/a5b44ff831d659e3772f/F5.jpg", "media/a5b44ff831d659e3772f/F6.jpg", "media/a5b44ff831d659e3772f/F7.jpg", "media/a5b44ff831d659e3772f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (white minivan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2de2ed5cce93e6e62be29bb95a5ca76ef8dbe854b03cb9913ce6b1f03caa0f25"}
{"qa_id": "MOSE__d262b7e4__object_world_motion_direction__01", "video_id": "MOSE__d262b7e4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (black compact suv) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/5985187d23759141a0af/F0.jpg", "media/5985187d23759141a0af/F1.jpg", "media/5985187d23759141a0af/F2.jpg", "media/5985187d23759141a0af/F3.jpg", "media/5985187d23759141a0af/F4.jpg", "media/5985187d23759141a0af/F5.jpg", "media/5985187d23759141a0af/F6.jpg", "media/5985187d23759141a0af/F7.jpg", "media/5985187d23759141a0af/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (black compact suv) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "826bd6acb42cf660214e69bfe037ba0b4011d7258d35958b04576b4422675d1e"}
{"qa_id": "MOSE__d262b7e4__camera_object_distance_change__02", "video_id": "MOSE__d262b7e4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object B (black compact suv) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5985187d23759141a0af/F0.jpg", "media/5985187d23759141a0af/F1.jpg", "media/5985187d23759141a0af/F2.jpg", "media/5985187d23759141a0af/F3.jpg", "media/5985187d23759141a0af/F4.jpg", "media/5985187d23759141a0af/F5.jpg", "media/5985187d23759141a0af/F6.jpg", "media/5985187d23759141a0af/F7.jpg", "media/5985187d23759141a0af/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object B (black compact suv) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3807c2532db85180de35aa451de5a57004322636ddfbde85663458fca369e8e7"}
{"qa_id": "MOSE__d262b7e4__object_pair_distance_change__03", "video_id": "MOSE__d262b7e4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (white minivan) and object B (black compact suv) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/34c3ffeab1fffbd8e350/F0.jpg", "media/34c3ffeab1fffbd8e350/F1.jpg", "media/34c3ffeab1fffbd8e350/F2.jpg", "media/34c3ffeab1fffbd8e350/F3.jpg", "media/34c3ffeab1fffbd8e350/F4.jpg", "media/34c3ffeab1fffbd8e350/F5.jpg", "media/34c3ffeab1fffbd8e350/F6.jpg", "media/34c3ffeab1fffbd8e350/F7.jpg", "media/34c3ffeab1fffbd8e350/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (white minivan) and object B (black compact suv) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eba73b7095b61ee7100306d6e57448fba399826915ec512d7735811e98d1fb94"}
{"qa_id": "MOSE__d29991b4__object_world_motion_direction__00", "video_id": "MOSE__d29991b4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in blue shirt and black shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "backward", "forward"], "frame_paths": ["media/9a14e2becc421841f472/F0.jpg", "media/9a14e2becc421841f472/F1.jpg", "media/9a14e2becc421841f472/F2.jpg", "media/9a14e2becc421841f472/F3.jpg", "media/9a14e2becc421841f472/F4.jpg", "media/9a14e2becc421841f472/F5.jpg", "media/9a14e2becc421841f472/F6.jpg", "media/9a14e2becc421841f472/F7.jpg", "media/9a14e2becc421841f472/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in blue shirt and black shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16ae760d57d2420fd26412eba34b7bfcd0a0ba37fe60616afbb881fe20a7ff80"}
{"qa_id": "MOSE__d29991b4__camera_object_distance_change__01", "video_id": "MOSE__d29991b4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (person in blue shirt and black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9a14e2becc421841f472/F0.jpg", "media/9a14e2becc421841f472/F1.jpg", "media/9a14e2becc421841f472/F2.jpg", "media/9a14e2becc421841f472/F3.jpg", "media/9a14e2becc421841f472/F4.jpg", "media/9a14e2becc421841f472/F5.jpg", "media/9a14e2becc421841f472/F6.jpg", "media/9a14e2becc421841f472/F7.jpg", "media/9a14e2becc421841f472/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (person in blue shirt and black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e40e7509bb033ed174ab5bbaaeb437bb1ada3cd408a83dbc54d6600f6680168"}
{"qa_id": "MOSE__d29991b4__object_pair_distance_change__02", "video_id": "MOSE__d29991b4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did the 3D distance between object A (person in blue shirt and black shorts) and object B (person in red shirt and green shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/0e2999fb8d0d24dbf0b2/F0.jpg", "media/0e2999fb8d0d24dbf0b2/F1.jpg", "media/0e2999fb8d0d24dbf0b2/F2.jpg", "media/0e2999fb8d0d24dbf0b2/F3.jpg", "media/0e2999fb8d0d24dbf0b2/F4.jpg", "media/0e2999fb8d0d24dbf0b2/F5.jpg", "media/0e2999fb8d0d24dbf0b2/F6.jpg", "media/0e2999fb8d0d24dbf0b2/F7.jpg", "media/0e2999fb8d0d24dbf0b2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (person in blue shirt and black shorts) and object B (person in red shirt and green shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dcebfbd8016cd7e2b4eca8fbadbe96294b1b9fbb5abe2a0baa588261616658ee"}
{"qa_id": "MOSE__d29991b4__object_travel_distance_comparison__03", "video_id": "MOSE__d29991b4", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and black shorts) or B (person in red shirt and green shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/0e2999fb8d0d24dbf0b2/F0.jpg", "media/0e2999fb8d0d24dbf0b2/F1.jpg", "media/0e2999fb8d0d24dbf0b2/F2.jpg", "media/0e2999fb8d0d24dbf0b2/F3.jpg", "media/0e2999fb8d0d24dbf0b2/F4.jpg", "media/0e2999fb8d0d24dbf0b2/F5.jpg", "media/0e2999fb8d0d24dbf0b2/F6.jpg", "media/0e2999fb8d0d24dbf0b2/F7.jpg", "media/0e2999fb8d0d24dbf0b2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and black shorts) or B (person in red shirt and green shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62019f96ce69bb7d7daf134de2e2288c4fc4f30117a7fcc63f3621224611151a"}
{"qa_id": "MOSE__d2aa8c66__camera_view_direction_change__00", "video_id": "MOSE__d2aa8c66", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/293019bd2569729674a6/F0.jpg", "media/293019bd2569729674a6/F1.jpg", "media/293019bd2569729674a6/F2.jpg", "media/293019bd2569729674a6/F3.jpg", "media/293019bd2569729674a6/F4.jpg", "media/293019bd2569729674a6/F5.jpg", "media/293019bd2569729674a6/F6.jpg", "media/293019bd2569729674a6/F7.jpg", "media/293019bd2569729674a6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6e894bc1bf0ae5cd4ff6634e46928a156052b6c353aca1b723fe2310a012edcd"}
{"qa_id": "MOSE__d2aa8c66__object_world_motion_direction__01", "video_id": "MOSE__d2aa8c66", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object B (white horse with a dark mane) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/6ed189266ae89f799ffd/F0.jpg", "media/6ed189266ae89f799ffd/F1.jpg", "media/6ed189266ae89f799ffd/F2.jpg", "media/6ed189266ae89f799ffd/F3.jpg", "media/6ed189266ae89f799ffd/F4.jpg", "media/6ed189266ae89f799ffd/F5.jpg", "media/6ed189266ae89f799ffd/F6.jpg", "media/6ed189266ae89f799ffd/F7.jpg", "media/6ed189266ae89f799ffd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object B (white horse with a dark mane) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40412b8807baff27b1c58a09da777e51ed9e518229cdbb0e8865e042c6c06ca8"}
{"qa_id": "MOSE__d2e75e95__object_pair_distance_change__00", "video_id": "MOSE__d2e75e95", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (man in dark blue long-sleeve shirt holding a stick) and object B (man in light blue t-shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/39c5473d18b6a3e8dff0/F0.jpg", "media/39c5473d18b6a3e8dff0/F1.jpg", "media/39c5473d18b6a3e8dff0/F2.jpg", "media/39c5473d18b6a3e8dff0/F3.jpg", "media/39c5473d18b6a3e8dff0/F4.jpg", "media/39c5473d18b6a3e8dff0/F5.jpg", "media/39c5473d18b6a3e8dff0/F6.jpg", "media/39c5473d18b6a3e8dff0/F7.jpg", "media/39c5473d18b6a3e8dff0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (man in dark blue long-sleeve shirt holding a stick) and object B (man in light blue t-shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "580186a2feb1c9389663fe85574a63db6cf46d36815d3d1a132bfa31be18bcd5"}
{"qa_id": "MOSE__d2f4b9e2__camera_object_distance_change__00", "video_id": "MOSE__d2f4b9e2", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object B (person in pink shirt riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2d833fa6aa4840a4990e/F0.jpg", "media/2d833fa6aa4840a4990e/F1.jpg", "media/2d833fa6aa4840a4990e/F2.jpg", "media/2d833fa6aa4840a4990e/F3.jpg", "media/2d833fa6aa4840a4990e/F4.jpg", "media/2d833fa6aa4840a4990e/F5.jpg", "media/2d833fa6aa4840a4990e/F6.jpg", "media/2d833fa6aa4840a4990e/F7.jpg", "media/2d833fa6aa4840a4990e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object B (person in pink shirt riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0932d37da9cb419167390d107afdb059e2286b3ff3c7129a28434d206e01bb8f"}
{"qa_id": "MOSE__d2f4b9e2__object_pair_distance_change__01", "video_id": "MOSE__d2f4b9e2", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (orange and white city bus) and object B (person in pink shirt riding a scooter) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b226d03245f4c011de3a/F0.jpg", "media/b226d03245f4c011de3a/F1.jpg", "media/b226d03245f4c011de3a/F2.jpg", "media/b226d03245f4c011de3a/F3.jpg", "media/b226d03245f4c011de3a/F4.jpg", "media/b226d03245f4c011de3a/F5.jpg", "media/b226d03245f4c011de3a/F6.jpg", "media/b226d03245f4c011de3a/F7.jpg", "media/b226d03245f4c011de3a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (orange and white city bus) and object B (person in pink shirt riding a scooter) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9d207cdc4844515c1a09084f906bef45da651e70ed818d9278fb1ffa51ad895a"}
{"qa_id": "MOSE__d2f4b9e2__object_travel_distance_comparison__02", "video_id": "MOSE__d2f4b9e2", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange and white city bus) or B (person in pink shirt riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/b226d03245f4c011de3a/F0.jpg", "media/b226d03245f4c011de3a/F1.jpg", "media/b226d03245f4c011de3a/F2.jpg", "media/b226d03245f4c011de3a/F3.jpg", "media/b226d03245f4c011de3a/F4.jpg", "media/b226d03245f4c011de3a/F5.jpg", "media/b226d03245f4c011de3a/F6.jpg", "media/b226d03245f4c011de3a/F7.jpg", "media/b226d03245f4c011de3a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange and white city bus) or B (person in pink shirt riding a scooter)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0eba1c90d6014f077982eeacc8b80221944602a8486aefc9a344fb2e560b4ceb"}
{"qa_id": "MOSE__d321dde4__camera_translation_direction__00", "video_id": "MOSE__d321dde4", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "up", "backward", "down"], "frame_paths": ["media/ade0a3610f33ab7d4423/F0.jpg", "media/ade0a3610f33ab7d4423/F1.jpg", "media/ade0a3610f33ab7d4423/F2.jpg", "media/ade0a3610f33ab7d4423/F3.jpg", "media/ade0a3610f33ab7d4423/F4.jpg", "media/ade0a3610f33ab7d4423/F5.jpg", "media/ade0a3610f33ab7d4423/F6.jpg", "media/ade0a3610f33ab7d4423/F7.jpg", "media/ade0a3610f33ab7d4423/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cd8529c1ed708e778e55ae7fd45ddf5b6da4fa4e1e13cf3fd361d5f1e798569d"}
{"qa_id": "MOSE__d321dde4__camera_view_direction_change__01", "video_id": "MOSE__d321dde4", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/ade0a3610f33ab7d4423/F0.jpg", "media/ade0a3610f33ab7d4423/F1.jpg", "media/ade0a3610f33ab7d4423/F2.jpg", "media/ade0a3610f33ab7d4423/F3.jpg", "media/ade0a3610f33ab7d4423/F4.jpg", "media/ade0a3610f33ab7d4423/F5.jpg", "media/ade0a3610f33ab7d4423/F6.jpg", "media/ade0a3610f33ab7d4423/F7.jpg", "media/ade0a3610f33ab7d4423/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55c282928f8940355d2075dba86d8619c91a164d28fb231d532d13fe514a5de1"}
{"qa_id": "MOSE__d321dde4__object_pair_distance_change__02", "video_id": "MOSE__d321dde4", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did the 3D distance between object A (light brown goat) and object B (brown goat) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7d2b88f516e6f03e959c/F0.jpg", "media/7d2b88f516e6f03e959c/F1.jpg", "media/7d2b88f516e6f03e959c/F2.jpg", "media/7d2b88f516e6f03e959c/F3.jpg", "media/7d2b88f516e6f03e959c/F4.jpg", "media/7d2b88f516e6f03e959c/F5.jpg", "media/7d2b88f516e6f03e959c/F6.jpg", "media/7d2b88f516e6f03e959c/F7.jpg", "media/7d2b88f516e6f03e959c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (light brown goat) and object B (brown goat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cbe25f9bcf36fd7e0c2b8ba517fe2a2a6dbd1783afdf17065a509caa90865453"}
{"qa_id": "MOSE__d321dde4__object_travel_distance_comparison__03", "video_id": "MOSE__d321dde4", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown goat) or B (brown goat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/7d2b88f516e6f03e959c/F0.jpg", "media/7d2b88f516e6f03e959c/F1.jpg", "media/7d2b88f516e6f03e959c/F2.jpg", "media/7d2b88f516e6f03e959c/F3.jpg", "media/7d2b88f516e6f03e959c/F4.jpg", "media/7d2b88f516e6f03e959c/F5.jpg", "media/7d2b88f516e6f03e959c/F6.jpg", "media/7d2b88f516e6f03e959c/F7.jpg", "media/7d2b88f516e6f03e959c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown goat) or B (brown goat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8cc0b93a774d8e6fac74ffc3c1389501db9aa81a47bdf9491dd39eea7b2a677"}
{"qa_id": "MOSE__d37e29a7__object_world_motion_direction__00", "video_id": "MOSE__d37e29a7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (light brown bull with large horns) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "left", "forward", "backward"], "frame_paths": ["media/4842759a39bd1cd92cba/F0.jpg", "media/4842759a39bd1cd92cba/F1.jpg", "media/4842759a39bd1cd92cba/F2.jpg", "media/4842759a39bd1cd92cba/F3.jpg", "media/4842759a39bd1cd92cba/F4.jpg", "media/4842759a39bd1cd92cba/F5.jpg", "media/4842759a39bd1cd92cba/F6.jpg", "media/4842759a39bd1cd92cba/F7.jpg", "media/4842759a39bd1cd92cba/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (light brown bull with large horns) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e5700d49bc144533c22b7457fbfd93b9f1c82b77476ca3d86751918f2f55a20"}
{"qa_id": "MOSE__d7058a0d__object_travel_distance_comparison__00", "video_id": "MOSE__d7058a0d", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey plaid suit) or B (person in white shirt holding red roses)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/9bd11858e5bc35c735f0/F0.jpg", "media/9bd11858e5bc35c735f0/F1.jpg", "media/9bd11858e5bc35c735f0/F2.jpg", "media/9bd11858e5bc35c735f0/F3.jpg", "media/9bd11858e5bc35c735f0/F4.jpg", "media/9bd11858e5bc35c735f0/F5.jpg", "media/9bd11858e5bc35c735f0/F6.jpg", "media/9bd11858e5bc35c735f0/F7.jpg", "media/9bd11858e5bc35c735f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey plaid suit) or B (person in white shirt holding red roses)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c6f3d6b3f87d78ee744ebf690fe8b0f1e89473b3260c84efad61336ee5f5fe06"}
{"qa_id": "MOSE__d7969c29__camera_view_direction_change__00", "video_id": "MOSE__d7969c29", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/b6c946ab66cb7df3c6f1/F0.jpg", "media/b6c946ab66cb7df3c6f1/F1.jpg", "media/b6c946ab66cb7df3c6f1/F2.jpg", "media/b6c946ab66cb7df3c6f1/F3.jpg", "media/b6c946ab66cb7df3c6f1/F4.jpg", "media/b6c946ab66cb7df3c6f1/F5.jpg", "media/b6c946ab66cb7df3c6f1/F6.jpg", "media/b6c946ab66cb7df3c6f1/F7.jpg", "media/b6c946ab66cb7df3c6f1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "MOSE__d904c048__object_world_motion_direction__00", "video_id": "MOSE__d904c048", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (orange fish with a rounded body and prominent eye) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/fc60db8397dde7401a08/F0.jpg", "media/fc60db8397dde7401a08/F1.jpg", "media/fc60db8397dde7401a08/F2.jpg", "media/fc60db8397dde7401a08/F3.jpg", "media/fc60db8397dde7401a08/F4.jpg", "media/fc60db8397dde7401a08/F5.jpg", "media/fc60db8397dde7401a08/F6.jpg", "media/fc60db8397dde7401a08/F7.jpg", "media/fc60db8397dde7401a08/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (orange fish with a rounded body and prominent eye) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d360f398fc716285bd7999fdf1d5971cfc3831ed216a4b695ff37f0e34bd9ba"}
{"qa_id": "MOSE__d904c048__object_pair_distance_change__01", "video_id": "MOSE__d904c048", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F5, did the 3D distance between object A (orange fish with a rounded body and prominent eye) and object B (small yellowish-orange fish near the gravel substrate) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/f3745779788f2128ce8e/F0.jpg", "media/f3745779788f2128ce8e/F1.jpg", "media/f3745779788f2128ce8e/F2.jpg", "media/f3745779788f2128ce8e/F3.jpg", "media/f3745779788f2128ce8e/F4.jpg", "media/f3745779788f2128ce8e/F5.jpg", "media/f3745779788f2128ce8e/F6.jpg", "media/f3745779788f2128ce8e/F7.jpg", "media/f3745779788f2128ce8e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did the 3D distance between object A (orange fish with a rounded body and prominent eye) and object B (small yellowish-orange fish near the gravel substrate) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5cc92da91717188af0eaf646204effb987ddc3d46434b42cb339d14e8328adee"}
{"qa_id": "MOSE__d904c048__object_travel_distance_comparison__02", "video_id": "MOSE__d904c048", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish with a rounded body and prominent eye) or B (small yellowish-orange fish near the gravel substrate)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f3745779788f2128ce8e/F0.jpg", "media/f3745779788f2128ce8e/F1.jpg", "media/f3745779788f2128ce8e/F2.jpg", "media/f3745779788f2128ce8e/F3.jpg", "media/f3745779788f2128ce8e/F4.jpg", "media/f3745779788f2128ce8e/F5.jpg", "media/f3745779788f2128ce8e/F6.jpg", "media/f3745779788f2128ce8e/F7.jpg", "media/f3745779788f2128ce8e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish with a rounded body and prominent eye) or B (small yellowish-orange fish near the gravel substrate)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "972550c0e5451d36437944cd994353cd5fb85813f77aa331f38e9be1b7e7404a"}
{"qa_id": "MOSE__da5e4bc5__object_world_motion_direction__00", "video_id": "MOSE__da5e4bc5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (elephant lying down in the foreground) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "backward", "left", "right"], "frame_paths": ["media/f9aecdf697d512427e08/F0.jpg", "media/f9aecdf697d512427e08/F1.jpg", "media/f9aecdf697d512427e08/F2.jpg", "media/f9aecdf697d512427e08/F3.jpg", "media/f9aecdf697d512427e08/F4.jpg", "media/f9aecdf697d512427e08/F5.jpg", "media/f9aecdf697d512427e08/F6.jpg", "media/f9aecdf697d512427e08/F7.jpg", "media/f9aecdf697d512427e08/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (elephant lying down in the foreground) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05edbcee10fa5746b3533cfd26b38c47b4d5ffd810194a34432f8f90f301013c"}
{"qa_id": "MOSE__da5e4bc5__object_world_motion_direction__01", "video_id": "MOSE__da5e4bc5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (elephant standing on the right side) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "left", "forward", "backward"], "frame_paths": ["media/b0da6b6015e480bffd72/F0.jpg", "media/b0da6b6015e480bffd72/F1.jpg", "media/b0da6b6015e480bffd72/F2.jpg", "media/b0da6b6015e480bffd72/F3.jpg", "media/b0da6b6015e480bffd72/F4.jpg", "media/b0da6b6015e480bffd72/F5.jpg", "media/b0da6b6015e480bffd72/F6.jpg", "media/b0da6b6015e480bffd72/F7.jpg", "media/b0da6b6015e480bffd72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (elephant standing on the right side) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13cd96522082ca663ba7e8d39190895e0933427b4d7257a9390713831084da91"}
{"qa_id": "MOSE__da5e4bc5__object_pair_distance_change__02", "video_id": "MOSE__da5e4bc5", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (elephant lying down in the foreground) and object B (elephant standing on the right side) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/4c05ab1c01afab252fbc/F0.jpg", "media/4c05ab1c01afab252fbc/F1.jpg", "media/4c05ab1c01afab252fbc/F2.jpg", "media/4c05ab1c01afab252fbc/F3.jpg", "media/4c05ab1c01afab252fbc/F4.jpg", "media/4c05ab1c01afab252fbc/F5.jpg", "media/4c05ab1c01afab252fbc/F6.jpg", "media/4c05ab1c01afab252fbc/F7.jpg", "media/4c05ab1c01afab252fbc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (elephant lying down in the foreground) and object B (elephant standing on the right side) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db510d86ec869b3f7941d9d0ccdd473787a17a6b26bafac913a84a807801a1e8"}
{"qa_id": "MOSE__dae93f5c__camera_translation_direction__00", "video_id": "MOSE__dae93f5c", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "forward", "up", "down"], "frame_paths": ["media/fbad160aa67e5b464019/F0.jpg", "media/fbad160aa67e5b464019/F1.jpg", "media/fbad160aa67e5b464019/F2.jpg", "media/fbad160aa67e5b464019/F3.jpg", "media/fbad160aa67e5b464019/F4.jpg", "media/fbad160aa67e5b464019/F5.jpg", "media/fbad160aa67e5b464019/F6.jpg", "media/fbad160aa67e5b464019/F7.jpg", "media/fbad160aa67e5b464019/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7031d1824d858774aca3b934b984590e065844a87ff88dddb51d36a90239db4f"}
{"qa_id": "MOSE__dae93f5c__object_world_motion_direction__01", "video_id": "MOSE__dae93f5c", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (orange fish located in the upper right area of the tank) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "up", "backward", "left"], "frame_paths": ["media/6765ece7bdfe07efd414/F0.jpg", "media/6765ece7bdfe07efd414/F1.jpg", "media/6765ece7bdfe07efd414/F2.jpg", "media/6765ece7bdfe07efd414/F3.jpg", "media/6765ece7bdfe07efd414/F4.jpg", "media/6765ece7bdfe07efd414/F5.jpg", "media/6765ece7bdfe07efd414/F6.jpg", "media/6765ece7bdfe07efd414/F7.jpg", "media/6765ece7bdfe07efd414/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (orange fish located in the upper right area of the tank) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "57fabc1553c457875781fdd26cf78481884768db8c6c63bb156394dc25f0064e"}
{"qa_id": "MOSE__dae93f5c__object_travel_distance_comparison__02", "video_id": "MOSE__dae93f5c", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish with a prominent head bump, positioned near the bottom gravel) or B (orange fish located in the upper right area of the tank)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/355ae1924650222926f9/F0.jpg", "media/355ae1924650222926f9/F1.jpg", "media/355ae1924650222926f9/F2.jpg", "media/355ae1924650222926f9/F3.jpg", "media/355ae1924650222926f9/F4.jpg", "media/355ae1924650222926f9/F5.jpg", "media/355ae1924650222926f9/F6.jpg", "media/355ae1924650222926f9/F7.jpg", "media/355ae1924650222926f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange fish with a prominent head bump, positioned near the bottom gravel) or B (orange fish located in the upper right area of the tank)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fde22ad2d1fa531e8824d2f270a98433d4f2b9bfac9d02d14a817e808e688de2"}
{"qa_id": "MOSE__dafeb75e__camera_object_distance_change__00", "video_id": "MOSE__dafeb75e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F6, did object A (dark-shelled turtle with patterned skin on the head and flippers) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/94d8317f7e134f43860a/F0.jpg", "media/94d8317f7e134f43860a/F1.jpg", "media/94d8317f7e134f43860a/F2.jpg", "media/94d8317f7e134f43860a/F3.jpg", "media/94d8317f7e134f43860a/F4.jpg", "media/94d8317f7e134f43860a/F5.jpg", "media/94d8317f7e134f43860a/F6.jpg", "media/94d8317f7e134f43860a/F7.jpg", "media/94d8317f7e134f43860a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (dark-shelled turtle with patterned skin on the head and flippers) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d583381d8d6cc22a2425fe305724e5e7049300def9af1f994db36d3c099e531"}
{"qa_id": "MOSE__dafeb75e__object_world_motion_direction__01", "video_id": "MOSE__dafeb75e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object B (dark-shelled turtle swimming in the background) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "backward", "left", "down", "forward"], "frame_paths": ["media/73baeb26d4e860d6fc5a/F0.jpg", "media/73baeb26d4e860d6fc5a/F1.jpg", "media/73baeb26d4e860d6fc5a/F2.jpg", "media/73baeb26d4e860d6fc5a/F3.jpg", "media/73baeb26d4e860d6fc5a/F4.jpg", "media/73baeb26d4e860d6fc5a/F5.jpg", "media/73baeb26d4e860d6fc5a/F6.jpg", "media/73baeb26d4e860d6fc5a/F7.jpg", "media/73baeb26d4e860d6fc5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object B (dark-shelled turtle swimming in the background) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. backward\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bfc8ed5972c434edfe4599df5af7ef312786f4d590bc7e5f0e01ccf323134812"}
{"qa_id": "MOSE__dafeb75e__object_pair_distance_change__02", "video_id": "MOSE__dafeb75e", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F6, did the 3D distance between object A (dark-shelled turtle with patterned skin on the head and flippers) and object B (dark-shelled turtle swimming in the background) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/2584da703cbb0a04f058/F0.jpg", "media/2584da703cbb0a04f058/F1.jpg", "media/2584da703cbb0a04f058/F2.jpg", "media/2584da703cbb0a04f058/F3.jpg", "media/2584da703cbb0a04f058/F4.jpg", "media/2584da703cbb0a04f058/F5.jpg", "media/2584da703cbb0a04f058/F6.jpg", "media/2584da703cbb0a04f058/F7.jpg", "media/2584da703cbb0a04f058/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did the 3D distance between object A (dark-shelled turtle with patterned skin on the head and flippers) and object B (dark-shelled turtle swimming in the background) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a714564e0f4ed6eef212470e50767390b5add748fdc35099490daa4f09a3bd8"}
{"qa_id": "MOSE__dbbbb401__camera_translation_direction__00", "video_id": "MOSE__dbbbb401", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "forward", "right", "backward"], "frame_paths": ["media/3f72d1523038baa44525/F0.jpg", "media/3f72d1523038baa44525/F1.jpg", "media/3f72d1523038baa44525/F2.jpg", "media/3f72d1523038baa44525/F3.jpg", "media/3f72d1523038baa44525/F4.jpg", "media/3f72d1523038baa44525/F5.jpg", "media/3f72d1523038baa44525/F6.jpg", "media/3f72d1523038baa44525/F7.jpg", "media/3f72d1523038baa44525/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "22b8953f9751b302687a317da1581ef2538b53698e8716a7f1b9ec2a8b7c1048"}
{"qa_id": "MOSE__dbbbb401__camera_object_distance_change__01", "video_id": "MOSE__dbbbb401", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, did object A (brown cow walking on the path) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/68f5709d0dd4a22593e5/F0.jpg", "media/68f5709d0dd4a22593e5/F1.jpg", "media/68f5709d0dd4a22593e5/F2.jpg", "media/68f5709d0dd4a22593e5/F3.jpg", "media/68f5709d0dd4a22593e5/F4.jpg", "media/68f5709d0dd4a22593e5/F5.jpg", "media/68f5709d0dd4a22593e5/F6.jpg", "media/68f5709d0dd4a22593e5/F7.jpg", "media/68f5709d0dd4a22593e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did object A (brown cow walking on the path) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fcbf49b4255d7a1fd3e9804676d4fd5766da53bb67a4c8ab9b0b6520fa08d5f"}
{"qa_id": "MOSE__dbbbb401__object_world_motion_direction__02", "video_id": "MOSE__dbbbb401", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object B (black cow standing on the left) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "left", "forward", "down"], "frame_paths": ["media/feb278ae142d82a7c218/F0.jpg", "media/feb278ae142d82a7c218/F1.jpg", "media/feb278ae142d82a7c218/F2.jpg", "media/feb278ae142d82a7c218/F3.jpg", "media/feb278ae142d82a7c218/F4.jpg", "media/feb278ae142d82a7c218/F5.jpg", "media/feb278ae142d82a7c218/F6.jpg", "media/feb278ae142d82a7c218/F7.jpg", "media/feb278ae142d82a7c218/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object B (black cow standing on the left) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fcae8c71610adde69962302faf2fec10bac78c719b60c819be772d58c3ad75ff"}
{"qa_id": "MOSE__dbbbb401__object_pair_distance_change__03", "video_id": "MOSE__dbbbb401", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, did the 3D distance between object A (brown cow walking on the path) and object B (black cow standing on the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b37eea980367ebeb0971/F0.jpg", "media/b37eea980367ebeb0971/F1.jpg", "media/b37eea980367ebeb0971/F2.jpg", "media/b37eea980367ebeb0971/F3.jpg", "media/b37eea980367ebeb0971/F4.jpg", "media/b37eea980367ebeb0971/F5.jpg", "media/b37eea980367ebeb0971/F6.jpg", "media/b37eea980367ebeb0971/F7.jpg", "media/b37eea980367ebeb0971/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did the 3D distance between object A (brown cow walking on the path) and object B (black cow standing on the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "12808edbfa17741519be17a98fd6107f9810137494e96a802259dcf77c27ad8e"}
{"qa_id": "MOSE__dbbbb401__object_travel_distance_comparison__04", "video_id": "MOSE__dbbbb401", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown cow walking on the path) or B (black cow standing on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/b37eea980367ebeb0971/F0.jpg", "media/b37eea980367ebeb0971/F1.jpg", "media/b37eea980367ebeb0971/F2.jpg", "media/b37eea980367ebeb0971/F3.jpg", "media/b37eea980367ebeb0971/F4.jpg", "media/b37eea980367ebeb0971/F5.jpg", "media/b37eea980367ebeb0971/F6.jpg", "media/b37eea980367ebeb0971/F7.jpg", "media/b37eea980367ebeb0971/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown cow walking on the path) or B (black cow standing on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73a13a1e1d3654e8d2cb4087028eabfeb90f30275253c5d9ab965d40c65afc5e"}
{"qa_id": "MOSE__dc54aab2__object_world_motion_direction__00", "video_id": "MOSE__dc54aab2", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (black racing motorcycle with number 13, ridden by a person in a white and black suit) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/69186f56653daa96ed49/F0.jpg", "media/69186f56653daa96ed49/F1.jpg", "media/69186f56653daa96ed49/F2.jpg", "media/69186f56653daa96ed49/F3.jpg", "media/69186f56653daa96ed49/F4.jpg", "media/69186f56653daa96ed49/F5.jpg", "media/69186f56653daa96ed49/F6.jpg", "media/69186f56653daa96ed49/F7.jpg", "media/69186f56653daa96ed49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (black racing motorcycle with number 13, ridden by a person in a white and black suit) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2cff44a0263b8d5b5c63f5d02c667868ef7494ca44d7af181c66661645ebfcb"}
{"qa_id": "MOSE__dd5a4739__object_world_motion_direction__00", "video_id": "MOSE__dd5a4739", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (ram with curved horns) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/49b576e8330b8a425089/F0.jpg", "media/49b576e8330b8a425089/F1.jpg", "media/49b576e8330b8a425089/F2.jpg", "media/49b576e8330b8a425089/F3.jpg", "media/49b576e8330b8a425089/F4.jpg", "media/49b576e8330b8a425089/F5.jpg", "media/49b576e8330b8a425089/F6.jpg", "media/49b576e8330b8a425089/F7.jpg", "media/49b576e8330b8a425089/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (ram with curved horns) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd2cd79e64118d765d5dbf87fde7edc795c922d53e27e9539c20a54f55c00426"}
{"qa_id": "MOSE__ddbdfa32__object_world_motion_direction__00", "video_id": "MOSE__ddbdfa32", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "forward", "up", "down", "right"], "frame_paths": ["media/f79d225f4370e7c4d03b/F0.jpg", "media/f79d225f4370e7c4d03b/F1.jpg", "media/f79d225f4370e7c4d03b/F2.jpg", "media/f79d225f4370e7c4d03b/F3.jpg", "media/f79d225f4370e7c4d03b/F4.jpg", "media/f79d225f4370e7c4d03b/F5.jpg", "media/f79d225f4370e7c4d03b/F6.jpg", "media/f79d225f4370e7c4d03b/F7.jpg", "media/f79d225f4370e7c4d03b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. forward\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1053f953b73cd282332a6d121411b70223b17589f3fa6afdcd33f63df4f4507a"}
{"qa_id": "MOSE__dde85a9d__object_world_motion_direction__00", "video_id": "MOSE__dde85a9d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant with visible tusks walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "down", "up", "left"], "frame_paths": ["media/5398b8f577f2579cdc97/F0.jpg", "media/5398b8f577f2579cdc97/F1.jpg", "media/5398b8f577f2579cdc97/F2.jpg", "media/5398b8f577f2579cdc97/F3.jpg", "media/5398b8f577f2579cdc97/F4.jpg", "media/5398b8f577f2579cdc97/F5.jpg", "media/5398b8f577f2579cdc97/F6.jpg", "media/5398b8f577f2579cdc97/F7.jpg", "media/5398b8f577f2579cdc97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (large adult elephant with visible tusks walking on the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "298133e3b4571276005fe3fcd8b60eb6a4aacbc594217e8cf53ae98ea2f7d690"}
{"qa_id": "MOSE__e147ab26__object_pair_distance_change__00", "video_id": "MOSE__e147ab26", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (zebra standing in the grass) and object B (small brown canid standing in the grass) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/008ed68a3a7a2b2b0667/F0.jpg", "media/008ed68a3a7a2b2b0667/F1.jpg", "media/008ed68a3a7a2b2b0667/F2.jpg", "media/008ed68a3a7a2b2b0667/F3.jpg", "media/008ed68a3a7a2b2b0667/F4.jpg", "media/008ed68a3a7a2b2b0667/F5.jpg", "media/008ed68a3a7a2b2b0667/F6.jpg", "media/008ed68a3a7a2b2b0667/F7.jpg", "media/008ed68a3a7a2b2b0667/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (zebra standing in the grass) and object B (small brown canid standing in the grass) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e2d1a37a4756d988f7cc84658cbd31787b6e822008b73cd0c64af4515cb4756"}
{"qa_id": "MOSE__e18723c4__object_world_motion_direction__00", "video_id": "MOSE__e18723c4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (motorcyclist in white and black racing suit with red and black helmet, riding a black and yellow bike) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "left", "right"], "frame_paths": ["media/66eea049f9f28aa59933/F0.jpg", "media/66eea049f9f28aa59933/F1.jpg", "media/66eea049f9f28aa59933/F2.jpg", "media/66eea049f9f28aa59933/F3.jpg", "media/66eea049f9f28aa59933/F4.jpg", "media/66eea049f9f28aa59933/F5.jpg", "media/66eea049f9f28aa59933/F6.jpg", "media/66eea049f9f28aa59933/F7.jpg", "media/66eea049f9f28aa59933/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (motorcyclist in white and black racing suit with red and black helmet, riding a black and yellow bike) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbd96259addd2a32a3de34f0852b8ff6d9442765cd2149783c17941bec7dbb8e"}
{"qa_id": "MOSE__e1d9fb32__object_world_motion_direction__00", "video_id": "MOSE__e1d9fb32", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (tan puppy) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "down", "right", "backward"], "frame_paths": ["media/897f93055b2c99d4b240/F0.jpg", "media/897f93055b2c99d4b240/F1.jpg", "media/897f93055b2c99d4b240/F2.jpg", "media/897f93055b2c99d4b240/F3.jpg", "media/897f93055b2c99d4b240/F4.jpg", "media/897f93055b2c99d4b240/F5.jpg", "media/897f93055b2c99d4b240/F6.jpg", "media/897f93055b2c99d4b240/F7.jpg", "media/897f93055b2c99d4b240/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (tan puppy) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0f9ec79cbbdb03cfd5976f1a574bf931d7ac2e670d8e50b7c78fd345d9a2b558"}
{"qa_id": "MOSE__e35a8262__object_world_motion_direction__00", "video_id": "MOSE__e35a8262", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (motocross rider in orange jersey with number 8) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "right", "down"], "frame_paths": ["media/cc19410b08da3037d6e3/F0.jpg", "media/cc19410b08da3037d6e3/F1.jpg", "media/cc19410b08da3037d6e3/F2.jpg", "media/cc19410b08da3037d6e3/F3.jpg", "media/cc19410b08da3037d6e3/F4.jpg", "media/cc19410b08da3037d6e3/F5.jpg", "media/cc19410b08da3037d6e3/F6.jpg", "media/cc19410b08da3037d6e3/F7.jpg", "media/cc19410b08da3037d6e3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (motocross rider in orange jersey with number 8) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5e037b54194fbbc41d3ef3608f2f3fadf090a44d166663232f42c112bc24147"}
{"qa_id": "MOSE__e3dcd186__object_world_motion_direction__00", "video_id": "MOSE__e3dcd186", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (orange and white goldfish with yellow head) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "backward", "forward"], "frame_paths": ["media/7c8f671b4a84cf3816cb/F0.jpg", "media/7c8f671b4a84cf3816cb/F1.jpg", "media/7c8f671b4a84cf3816cb/F2.jpg", "media/7c8f671b4a84cf3816cb/F3.jpg", "media/7c8f671b4a84cf3816cb/F4.jpg", "media/7c8f671b4a84cf3816cb/F5.jpg", "media/7c8f671b4a84cf3816cb/F6.jpg", "media/7c8f671b4a84cf3816cb/F7.jpg", "media/7c8f671b4a84cf3816cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (orange and white goldfish with yellow head) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e2c163d335f8d2f2a5624bf610c1abd1b8cb562be863bc9f9411c21a8a52ed6d"}
{"qa_id": "MOSE__e3dcd186__object_pair_distance_change__01", "video_id": "MOSE__e3dcd186", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (orange and white goldfish) and object B (orange and white goldfish with yellow head) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/e2a85a513b662fd64a91/F0.jpg", "media/e2a85a513b662fd64a91/F1.jpg", "media/e2a85a513b662fd64a91/F2.jpg", "media/e2a85a513b662fd64a91/F3.jpg", "media/e2a85a513b662fd64a91/F4.jpg", "media/e2a85a513b662fd64a91/F5.jpg", "media/e2a85a513b662fd64a91/F6.jpg", "media/e2a85a513b662fd64a91/F7.jpg", "media/e2a85a513b662fd64a91/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (orange and white goldfish) and object B (orange and white goldfish with yellow head) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05bef0e80511aa4bbe798f8eb5918ab0234a139911f020a022c533d933b00282"}
{"qa_id": "MOSE__e495537a__camera_object_distance_change__00", "video_id": "MOSE__e495537a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object B (large dark barge in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8a0d5c723f299f353f6b/F0.jpg", "media/8a0d5c723f299f353f6b/F1.jpg", "media/8a0d5c723f299f353f6b/F2.jpg", "media/8a0d5c723f299f353f6b/F3.jpg", "media/8a0d5c723f299f353f6b/F4.jpg", "media/8a0d5c723f299f353f6b/F5.jpg", "media/8a0d5c723f299f353f6b/F6.jpg", "media/8a0d5c723f299f353f6b/F7.jpg", "media/8a0d5c723f299f353f6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object B (large dark barge in the foreground) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b47e7a24864e99dd6069f74de9e441f5724e4f355ce7427c9dff63a429e08f33"}
{"qa_id": "MOSE__e495537a__object_travel_distance_comparison__01", "video_id": "MOSE__e495537a", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange boat in the distance) or B (large dark barge in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/9f11175b9c04eabc507d/F0.jpg", "media/9f11175b9c04eabc507d/F1.jpg", "media/9f11175b9c04eabc507d/F2.jpg", "media/9f11175b9c04eabc507d/F3.jpg", "media/9f11175b9c04eabc507d/F4.jpg", "media/9f11175b9c04eabc507d/F5.jpg", "media/9f11175b9c04eabc507d/F6.jpg", "media/9f11175b9c04eabc507d/F7.jpg", "media/9f11175b9c04eabc507d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (orange boat in the distance) or B (large dark barge in the foreground)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f58626dc4d1fa7fde06b0fe4071390f4a4b0a864af7c9e6c7d11d8205fabaac"}
{"qa_id": "MOSE__e4a72496__camera_object_distance_change__00", "video_id": "MOSE__e4a72496", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object A (dark grey cat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/27635dd1e66b9e98a479/F0.jpg", "media/27635dd1e66b9e98a479/F1.jpg", "media/27635dd1e66b9e98a479/F2.jpg", "media/27635dd1e66b9e98a479/F3.jpg", "media/27635dd1e66b9e98a479/F4.jpg", "media/27635dd1e66b9e98a479/F5.jpg", "media/27635dd1e66b9e98a479/F6.jpg", "media/27635dd1e66b9e98a479/F7.jpg", "media/27635dd1e66b9e98a479/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (dark grey cat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "086598b41f90534375164517ac1b73be9a5b7658b8c544d311f1de2f349af76e"}
{"qa_id": "MOSE__e4a72496__camera_object_distance_change__01", "video_id": "MOSE__e4a72496", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object B (white cardboard box with text) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/75ad3f28029a9092f559/F0.jpg", "media/75ad3f28029a9092f559/F1.jpg", "media/75ad3f28029a9092f559/F2.jpg", "media/75ad3f28029a9092f559/F3.jpg", "media/75ad3f28029a9092f559/F4.jpg", "media/75ad3f28029a9092f559/F5.jpg", "media/75ad3f28029a9092f559/F6.jpg", "media/75ad3f28029a9092f559/F7.jpg", "media/75ad3f28029a9092f559/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object B (white cardboard box with text) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b1725cd4b012c7bd4f7337ec74163feb7069b3bf8339e7acef1b70ccdb1ed1e0"}
{"qa_id": "MOSE__e4f4ddae__camera_translation_direction__00", "video_id": "MOSE__e4f4ddae", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "down", "right", "backward"], "frame_paths": ["media/fbf9fc063d13adf8e607/F0.jpg", "media/fbf9fc063d13adf8e607/F1.jpg", "media/fbf9fc063d13adf8e607/F2.jpg", "media/fbf9fc063d13adf8e607/F3.jpg", "media/fbf9fc063d13adf8e607/F4.jpg", "media/fbf9fc063d13adf8e607/F5.jpg", "media/fbf9fc063d13adf8e607/F6.jpg", "media/fbf9fc063d13adf8e607/F7.jpg", "media/fbf9fc063d13adf8e607/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. down\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e013ebc6e4f160cbd7e5b74245d1ddc9a6a7139465d879c1df04c91433bd1c4f"}
{"qa_id": "MOSE__e5665030__object_world_motion_direction__00", "video_id": "MOSE__e5665030", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (tiger walking on the rocks near the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "up", "backward", "left"], "frame_paths": ["media/ae64c9e90961855cebc7/F0.jpg", "media/ae64c9e90961855cebc7/F1.jpg", "media/ae64c9e90961855cebc7/F2.jpg", "media/ae64c9e90961855cebc7/F3.jpg", "media/ae64c9e90961855cebc7/F4.jpg", "media/ae64c9e90961855cebc7/F5.jpg", "media/ae64c9e90961855cebc7/F6.jpg", "media/ae64c9e90961855cebc7/F7.jpg", "media/ae64c9e90961855cebc7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (tiger walking on the rocks near the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5102bcc4821540d15433f80e602a51b88c8b9f3afe9b6dea1eb45cbd48b7591a"}
{"qa_id": "MOSE__e5665030__object_pair_distance_change__01", "video_id": "MOSE__e5665030", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (tiger walking on the rocks near the water) and object B (tiger standing on the rocks near the water) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/e26e4a60651f2e3102c4/F0.jpg", "media/e26e4a60651f2e3102c4/F1.jpg", "media/e26e4a60651f2e3102c4/F2.jpg", "media/e26e4a60651f2e3102c4/F3.jpg", "media/e26e4a60651f2e3102c4/F4.jpg", "media/e26e4a60651f2e3102c4/F5.jpg", "media/e26e4a60651f2e3102c4/F6.jpg", "media/e26e4a60651f2e3102c4/F7.jpg", "media/e26e4a60651f2e3102c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (tiger walking on the rocks near the water) and object B (tiger standing on the rocks near the water) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fdc71b893e34354336939e01337aae7ac6c136df46bfa20fb8d15febafb84b0b"}
{"qa_id": "MOSE__e607450d__object_world_motion_direction__00", "video_id": "MOSE__e607450d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in orange shirt and blue jeans walking a dog) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "up", "forward", "down"], "frame_paths": ["media/fad379d5d79ebf08058b/F0.jpg", "media/fad379d5d79ebf08058b/F1.jpg", "media/fad379d5d79ebf08058b/F2.jpg", "media/fad379d5d79ebf08058b/F3.jpg", "media/fad379d5d79ebf08058b/F4.jpg", "media/fad379d5d79ebf08058b/F5.jpg", "media/fad379d5d79ebf08058b/F6.jpg", "media/fad379d5d79ebf08058b/F7.jpg", "media/fad379d5d79ebf08058b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in orange shirt and blue jeans walking a dog) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f99319f41cd878f1442143956dea70586b8774725138481b43eb86e27c7391ac"}
{"qa_id": "MOSE__e71f9faa__camera_object_distance_change__00", "video_id": "MOSE__e71f9faa", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b709bea1875aec39f8b9/F0.jpg", "media/b709bea1875aec39f8b9/F1.jpg", "media/b709bea1875aec39f8b9/F2.jpg", "media/b709bea1875aec39f8b9/F3.jpg", "media/b709bea1875aec39f8b9/F4.jpg", "media/b709bea1875aec39f8b9/F5.jpg", "media/b709bea1875aec39f8b9/F6.jpg", "media/b709bea1875aec39f8b9/F7.jpg", "media/b709bea1875aec39f8b9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (person in blue shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "395563998de9a7671a59b0081dafd9388b0e9b490406ae49a0185469d0ac629d"}
{"qa_id": "MOSE__e71f9faa__camera_object_distance_change__01", "video_id": "MOSE__e71f9faa", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object B (person in green shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c57bebefa23f5b101742/F0.jpg", "media/c57bebefa23f5b101742/F1.jpg", "media/c57bebefa23f5b101742/F2.jpg", "media/c57bebefa23f5b101742/F3.jpg", "media/c57bebefa23f5b101742/F4.jpg", "media/c57bebefa23f5b101742/F5.jpg", "media/c57bebefa23f5b101742/F6.jpg", "media/c57bebefa23f5b101742/F7.jpg", "media/c57bebefa23f5b101742/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object B (person in green shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1768fd0a208c145cafe950b77b7a338054aca3184161e80b4096f8012126df2a"}
{"qa_id": "MOSE__e7529cbc__object_world_motion_direction__00", "video_id": "MOSE__e7529cbc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (blue city bus) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "up", "backward", "forward"], "frame_paths": ["media/315f8c3b1c64a5d5145f/F0.jpg", "media/315f8c3b1c64a5d5145f/F1.jpg", "media/315f8c3b1c64a5d5145f/F2.jpg", "media/315f8c3b1c64a5d5145f/F3.jpg", "media/315f8c3b1c64a5d5145f/F4.jpg", "media/315f8c3b1c64a5d5145f/F5.jpg", "media/315f8c3b1c64a5d5145f/F6.jpg", "media/315f8c3b1c64a5d5145f/F7.jpg", "media/315f8c3b1c64a5d5145f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (blue city bus) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4fa79d13b4c4ed3be863d8530100205e3672b3adaba5087fb1802ee473fa0672"}
{"qa_id": "MOSE__e7529cbc__camera_object_distance_change__01", "video_id": "MOSE__e7529cbc", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (blue city bus) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/315f8c3b1c64a5d5145f/F0.jpg", "media/315f8c3b1c64a5d5145f/F1.jpg", "media/315f8c3b1c64a5d5145f/F2.jpg", "media/315f8c3b1c64a5d5145f/F3.jpg", "media/315f8c3b1c64a5d5145f/F4.jpg", "media/315f8c3b1c64a5d5145f/F5.jpg", "media/315f8c3b1c64a5d5145f/F6.jpg", "media/315f8c3b1c64a5d5145f/F7.jpg", "media/315f8c3b1c64a5d5145f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (blue city bus) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa33cd39d8655a254c2f0041d9672dd65023225218a9e06ac18b3e6a4b65b4fd"}
{"qa_id": "MOSE__e7b34ab6__object_world_motion_direction__00", "video_id": "MOSE__e7b34ab6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (white goose with orange beak) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "left", "up"], "frame_paths": ["media/de996aa6b169755b4b74/F0.jpg", "media/de996aa6b169755b4b74/F1.jpg", "media/de996aa6b169755b4b74/F2.jpg", "media/de996aa6b169755b4b74/F3.jpg", "media/de996aa6b169755b4b74/F4.jpg", "media/de996aa6b169755b4b74/F5.jpg", "media/de996aa6b169755b4b74/F6.jpg", "media/de996aa6b169755b4b74/F7.jpg", "media/de996aa6b169755b4b74/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (white goose with orange beak) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bae8cfa284fb61b4974324055b0bc74debfde2adc5b1fcba36dc06e250e6bc16"}
{"qa_id": "MOSE__e7b34ab6__object_pair_distance_change__01", "video_id": "MOSE__e7b34ab6", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (white goose with orange beak) and object B (grey goose viewed from behind) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/1a9117dd709b849afe3a/F0.jpg", "media/1a9117dd709b849afe3a/F1.jpg", "media/1a9117dd709b849afe3a/F2.jpg", "media/1a9117dd709b849afe3a/F3.jpg", "media/1a9117dd709b849afe3a/F4.jpg", "media/1a9117dd709b849afe3a/F5.jpg", "media/1a9117dd709b849afe3a/F6.jpg", "media/1a9117dd709b849afe3a/F7.jpg", "media/1a9117dd709b849afe3a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (white goose with orange beak) and object B (grey goose viewed from behind) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91a454cf5195c9a5adaa05ac155e0c9dae0abaff9b24c04c00cbe799cfd33092"}
{"qa_id": "MOSE__e99bd793__camera_object_distance_change__00", "video_id": "MOSE__e99bd793", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F5, did object A (person in teal shirt and grey shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/00387ae73b366063f4cf/F0.jpg", "media/00387ae73b366063f4cf/F1.jpg", "media/00387ae73b366063f4cf/F2.jpg", "media/00387ae73b366063f4cf/F3.jpg", "media/00387ae73b366063f4cf/F4.jpg", "media/00387ae73b366063f4cf/F5.jpg", "media/00387ae73b366063f4cf/F6.jpg", "media/00387ae73b366063f4cf/F7.jpg", "media/00387ae73b366063f4cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (person in teal shirt and grey shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e098b777dc9ef6b968511b0e3bcb6a9a55ac0bb468d3fb822f6a3d6a1b703a72"}
{"qa_id": "MOSE__eaff3301__camera_translation_direction__00", "video_id": "MOSE__eaff3301", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/e7addbd5c8ba8223ea4e/F0.jpg", "media/e7addbd5c8ba8223ea4e/F1.jpg", "media/e7addbd5c8ba8223ea4e/F2.jpg", "media/e7addbd5c8ba8223ea4e/F3.jpg", "media/e7addbd5c8ba8223ea4e/F4.jpg", "media/e7addbd5c8ba8223ea4e/F5.jpg", "media/e7addbd5c8ba8223ea4e/F6.jpg", "media/e7addbd5c8ba8223ea4e/F7.jpg", "media/e7addbd5c8ba8223ea4e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "147791865be245d59103fa73cd73b85c7ee704befea5ab8eb09648be4a9b86df"}
{"qa_id": "MOSE__eaff3301__camera_view_direction_change__01", "video_id": "MOSE__eaff3301", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/e7addbd5c8ba8223ea4e/F0.jpg", "media/e7addbd5c8ba8223ea4e/F1.jpg", "media/e7addbd5c8ba8223ea4e/F2.jpg", "media/e7addbd5c8ba8223ea4e/F3.jpg", "media/e7addbd5c8ba8223ea4e/F4.jpg", "media/e7addbd5c8ba8223ea4e/F5.jpg", "media/e7addbd5c8ba8223ea4e/F6.jpg", "media/e7addbd5c8ba8223ea4e/F7.jpg", "media/e7addbd5c8ba8223ea4e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "MOSE__ee059f99__object_world_motion_direction__00", "video_id": "MOSE__ee059f99", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (white goose with black spots) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "down", "forward", "right"], "frame_paths": ["media/cd3761da9e09d0003a64/F0.jpg", "media/cd3761da9e09d0003a64/F1.jpg", "media/cd3761da9e09d0003a64/F2.jpg", "media/cd3761da9e09d0003a64/F3.jpg", "media/cd3761da9e09d0003a64/F4.jpg", "media/cd3761da9e09d0003a64/F5.jpg", "media/cd3761da9e09d0003a64/F6.jpg", "media/cd3761da9e09d0003a64/F7.jpg", "media/cd3761da9e09d0003a64/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (white goose with black spots) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1778dacdc7e91fac9a1fb397b0d05d3ab5e4a0f5afafb628068d719a55a11af4"}
{"qa_id": "MOSE__ee059f99__object_pair_distance_change__01", "video_id": "MOSE__ee059f99", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (large grey and white goose with brown neck) and object B (white goose with black spots) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d6e1829740edeb71fce0/F0.jpg", "media/d6e1829740edeb71fce0/F1.jpg", "media/d6e1829740edeb71fce0/F2.jpg", "media/d6e1829740edeb71fce0/F3.jpg", "media/d6e1829740edeb71fce0/F4.jpg", "media/d6e1829740edeb71fce0/F5.jpg", "media/d6e1829740edeb71fce0/F6.jpg", "media/d6e1829740edeb71fce0/F7.jpg", "media/d6e1829740edeb71fce0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g6"}, "requested_object_ids": ["g4", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (large grey and white goose with brown neck) and object B (white goose with black spots) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9933ad78d879395df4a24e29b5096e95ab5db31b153c3d1faea8b299343e029d"}
{"qa_id": "MOSE__ee91f783__object_pair_distance_change__00", "video_id": "MOSE__ee91f783", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F2, did the 3D distance between object A (grey pigeon with black wing bars) and object B (grey pigeon with black wing bars) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/78b06154a2cab8ff88bb/F0.jpg", "media/78b06154a2cab8ff88bb/F1.jpg", "media/78b06154a2cab8ff88bb/F2.jpg", "media/78b06154a2cab8ff88bb/F3.jpg", "media/78b06154a2cab8ff88bb/F4.jpg", "media/78b06154a2cab8ff88bb/F5.jpg", "media/78b06154a2cab8ff88bb/F6.jpg", "media/78b06154a2cab8ff88bb/F7.jpg", "media/78b06154a2cab8ff88bb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (grey pigeon with black wing bars) and object B (grey pigeon with black wing bars) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dfd37f8c731df369345933458bb5fdca9144739b4feb9b87b1d0492d5ce3bdb5"}
{"qa_id": "MOSE__ee91f783__object_travel_distance_comparison__01", "video_id": "MOSE__ee91f783", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (grey pigeon with black wing bars) or B (grey pigeon with black wing bars)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/78b06154a2cab8ff88bb/F0.jpg", "media/78b06154a2cab8ff88bb/F1.jpg", "media/78b06154a2cab8ff88bb/F2.jpg", "media/78b06154a2cab8ff88bb/F3.jpg", "media/78b06154a2cab8ff88bb/F4.jpg", "media/78b06154a2cab8ff88bb/F5.jpg", "media/78b06154a2cab8ff88bb/F6.jpg", "media/78b06154a2cab8ff88bb/F7.jpg", "media/78b06154a2cab8ff88bb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (grey pigeon with black wing bars) or B (grey pigeon with black wing bars)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "610efeee1c4aeb747174a4e0f35467776988d8d85c53b485de33e1248f0f2856"}
{"qa_id": "MOSE__ee9caa47__camera_view_direction_change__00", "video_id": "MOSE__ee9caa47", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/9b4842acb2ed7ea8b937/F0.jpg", "media/9b4842acb2ed7ea8b937/F1.jpg", "media/9b4842acb2ed7ea8b937/F2.jpg", "media/9b4842acb2ed7ea8b937/F3.jpg", "media/9b4842acb2ed7ea8b937/F4.jpg", "media/9b4842acb2ed7ea8b937/F5.jpg", "media/9b4842acb2ed7ea8b937/F6.jpg", "media/9b4842acb2ed7ea8b937/F7.jpg", "media/9b4842acb2ed7ea8b937/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bec691eba368f26b6ba745fdd6a3eb3939004e6d5b9e30ac931bbc88d0b17629"}
{"qa_id": "MOSE__ee9caa47__camera_object_distance_change__01", "video_id": "MOSE__ee9caa47", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did object A (motorcycle ridden by a person in a blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a2cc657b3c837f1dafb8/F0.jpg", "media/a2cc657b3c837f1dafb8/F1.jpg", "media/a2cc657b3c837f1dafb8/F2.jpg", "media/a2cc657b3c837f1dafb8/F3.jpg", "media/a2cc657b3c837f1dafb8/F4.jpg", "media/a2cc657b3c837f1dafb8/F5.jpg", "media/a2cc657b3c837f1dafb8/F6.jpg", "media/a2cc657b3c837f1dafb8/F7.jpg", "media/a2cc657b3c837f1dafb8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (motorcycle ridden by a person in a blue shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa7d57f2a33c170455d76997e3bb336755b8510ee9fd26aa08ee0d1057c2053e"}
{"qa_id": "MOSE__ee9caa47__object_world_motion_direction__02", "video_id": "MOSE__ee9caa47", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object B (motorcycle ridden by a person in a dark shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "up", "backward", "left"], "frame_paths": ["media/538f07ce04eb6025ff00/F0.jpg", "media/538f07ce04eb6025ff00/F1.jpg", "media/538f07ce04eb6025ff00/F2.jpg", "media/538f07ce04eb6025ff00/F3.jpg", "media/538f07ce04eb6025ff00/F4.jpg", "media/538f07ce04eb6025ff00/F5.jpg", "media/538f07ce04eb6025ff00/F6.jpg", "media/538f07ce04eb6025ff00/F7.jpg", "media/538f07ce04eb6025ff00/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object B (motorcycle ridden by a person in a dark shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4b82a711dc7f8b5751649c9dfe18f567ee412111cf068a0221a6aaf1984942a7"}
{"qa_id": "MOSE__eeaaa67e__object_travel_distance_comparison__00", "video_id": "MOSE__eeaaa67e", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant standing in a grassy field) or B (person lying on the grass with legs extended)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/55186d3ad530c52c53c9/F0.jpg", "media/55186d3ad530c52c53c9/F1.jpg", "media/55186d3ad530c52c53c9/F2.jpg", "media/55186d3ad530c52c53c9/F3.jpg", "media/55186d3ad530c52c53c9/F4.jpg", "media/55186d3ad530c52c53c9/F5.jpg", "media/55186d3ad530c52c53c9/F6.jpg", "media/55186d3ad530c52c53c9/F7.jpg", "media/55186d3ad530c52c53c9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (large grey elephant standing in a grassy field) or B (person lying on the grass with legs extended)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c74d476b4e6b322fc48a8e633eb37f6cdd99e08adaac3be2007575c6ebc00878"}
{"qa_id": "MOSE__efa25913__object_world_motion_direction__00", "video_id": "MOSE__efa25913", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (elephant in the center with a rider) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/6e8ef8e4da277599181b/F0.jpg", "media/6e8ef8e4da277599181b/F1.jpg", "media/6e8ef8e4da277599181b/F2.jpg", "media/6e8ef8e4da277599181b/F3.jpg", "media/6e8ef8e4da277599181b/F4.jpg", "media/6e8ef8e4da277599181b/F5.jpg", "media/6e8ef8e4da277599181b/F6.jpg", "media/6e8ef8e4da277599181b/F7.jpg", "media/6e8ef8e4da277599181b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (elephant in the center with a rider) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9bfc994d335ceefa2b44dd4ec3e1b2b6b91a8d2554d6122de6154960be27361f"}
{"qa_id": "MOSE__efdab888__camera_view_direction_change__00", "video_id": "MOSE__efdab888", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/7097d0ced51e512f9deb/F0.jpg", "media/7097d0ced51e512f9deb/F1.jpg", "media/7097d0ced51e512f9deb/F2.jpg", "media/7097d0ced51e512f9deb/F3.jpg", "media/7097d0ced51e512f9deb/F4.jpg", "media/7097d0ced51e512f9deb/F5.jpg", "media/7097d0ced51e512f9deb/F6.jpg", "media/7097d0ced51e512f9deb/F7.jpg", "media/7097d0ced51e512f9deb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "76c9c3f5961d9d6f52825eb0d0e515757efff2935c5c5bdce255bc5e2350e01b"}
{"qa_id": "MOSE__efe9cbca__object_world_motion_direction__00", "video_id": "MOSE__efe9cbca", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (tiger walking in the background) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "forward", "up", "down", "backward"], "frame_paths": ["media/e165c49058c6b03be6c7/F0.jpg", "media/e165c49058c6b03be6c7/F1.jpg", "media/e165c49058c6b03be6c7/F2.jpg", "media/e165c49058c6b03be6c7/F3.jpg", "media/e165c49058c6b03be6c7/F4.jpg", "media/e165c49058c6b03be6c7/F5.jpg", "media/e165c49058c6b03be6c7/F6.jpg", "media/e165c49058c6b03be6c7/F7.jpg", "media/e165c49058c6b03be6c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (tiger walking in the background) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. forward\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3eb95568bb1780c7d1e9b24fb2548fce312c2c5b07c329953019d2e25bb6b174"}
{"qa_id": "MOSE__efe9cbca__object_pair_distance_change__01", "video_id": "MOSE__efe9cbca", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F2, did the 3D distance between object A (tiger walking in the background) and object B (tiger standing near the tree stumps) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/d8f92f826808b9886c0e/F0.jpg", "media/d8f92f826808b9886c0e/F1.jpg", "media/d8f92f826808b9886c0e/F2.jpg", "media/d8f92f826808b9886c0e/F3.jpg", "media/d8f92f826808b9886c0e/F4.jpg", "media/d8f92f826808b9886c0e/F5.jpg", "media/d8f92f826808b9886c0e/F6.jpg", "media/d8f92f826808b9886c0e/F7.jpg", "media/d8f92f826808b9886c0e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (tiger walking in the background) and object B (tiger standing near the tree stumps) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8e5819975c3e6fe278617de425686227c6a49283c04f4c8c8532daf2cf5b33d"}
{"qa_id": "MOSE__f0617f71__camera_object_distance_change__00", "video_id": "MOSE__f0617f71", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (white rabbit with pink ears) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/00fb55b649c24a5c7611/F0.jpg", "media/00fb55b649c24a5c7611/F1.jpg", "media/00fb55b649c24a5c7611/F2.jpg", "media/00fb55b649c24a5c7611/F3.jpg", "media/00fb55b649c24a5c7611/F4.jpg", "media/00fb55b649c24a5c7611/F5.jpg", "media/00fb55b649c24a5c7611/F6.jpg", "media/00fb55b649c24a5c7611/F7.jpg", "media/00fb55b649c24a5c7611/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (white rabbit with pink ears) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "21ac31b78f8aaef304fee337903dbb0c704dcc5c07c8cec37cad30596f612ebe"}
{"qa_id": "MOSE__f0617f71__object_travel_distance_comparison__01", "video_id": "MOSE__f0617f71", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white rabbit with pink ears) or B (black rabbit with long ears)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/360bb990df74d919eca6/F0.jpg", "media/360bb990df74d919eca6/F1.jpg", "media/360bb990df74d919eca6/F2.jpg", "media/360bb990df74d919eca6/F3.jpg", "media/360bb990df74d919eca6/F4.jpg", "media/360bb990df74d919eca6/F5.jpg", "media/360bb990df74d919eca6/F6.jpg", "media/360bb990df74d919eca6/F7.jpg", "media/360bb990df74d919eca6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white rabbit with pink ears) or B (black rabbit with long ears)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a518ffac693f5ae7a0f0c835fc7dc0154f090ab4be85d49a558e872b2d26bfd9"}
{"qa_id": "MOSE__f0b922bf__object_world_motion_direction__00", "video_id": "MOSE__f0b922bf", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (white scooter with rider in black jacket and white helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "right", "forward", "backward"], "frame_paths": ["media/c80351c8d206b0de7551/F0.jpg", "media/c80351c8d206b0de7551/F1.jpg", "media/c80351c8d206b0de7551/F2.jpg", "media/c80351c8d206b0de7551/F3.jpg", "media/c80351c8d206b0de7551/F4.jpg", "media/c80351c8d206b0de7551/F5.jpg", "media/c80351c8d206b0de7551/F6.jpg", "media/c80351c8d206b0de7551/F7.jpg", "media/c80351c8d206b0de7551/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (white scooter with rider in black jacket and white helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cfcc148fc1c16971d2f9cc86dca63921edc353a67b2305a165c7e038fa08ff6c"}
{"qa_id": "MOSE__f0b922bf__camera_object_distance_change__01", "video_id": "MOSE__f0b922bf", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (white scooter with rider in black jacket and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c80351c8d206b0de7551/F0.jpg", "media/c80351c8d206b0de7551/F1.jpg", "media/c80351c8d206b0de7551/F2.jpg", "media/c80351c8d206b0de7551/F3.jpg", "media/c80351c8d206b0de7551/F4.jpg", "media/c80351c8d206b0de7551/F5.jpg", "media/c80351c8d206b0de7551/F6.jpg", "media/c80351c8d206b0de7551/F7.jpg", "media/c80351c8d206b0de7551/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (white scooter with rider in black jacket and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3609e37d7d7c473033fd72c77a844a7fcb14d5d9a25b6cb179ae51fb8510554f"}
{"qa_id": "MOSE__f0b922bf__object_world_motion_direction__02", "video_id": "MOSE__f0b922bf", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (red scooter with rider in dark clothing and white helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "left", "up", "forward"], "frame_paths": ["media/8fd408e33c8258c5a211/F0.jpg", "media/8fd408e33c8258c5a211/F1.jpg", "media/8fd408e33c8258c5a211/F2.jpg", "media/8fd408e33c8258c5a211/F3.jpg", "media/8fd408e33c8258c5a211/F4.jpg", "media/8fd408e33c8258c5a211/F5.jpg", "media/8fd408e33c8258c5a211/F6.jpg", "media/8fd408e33c8258c5a211/F7.jpg", "media/8fd408e33c8258c5a211/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (red scooter with rider in dark clothing and white helmet) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b87f3a58707d824ff2fd8eb678539da36bbe0f19a829e22c92bbad31493f9eea"}
{"qa_id": "MOSE__f0b922bf__object_pair_distance_change__03", "video_id": "MOSE__f0b922bf", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (white scooter with rider in black jacket and white helmet) and object B (red scooter with rider in dark clothing and white helmet) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/50a4aeac90bf6bde5f26/F0.jpg", "media/50a4aeac90bf6bde5f26/F1.jpg", "media/50a4aeac90bf6bde5f26/F2.jpg", "media/50a4aeac90bf6bde5f26/F3.jpg", "media/50a4aeac90bf6bde5f26/F4.jpg", "media/50a4aeac90bf6bde5f26/F5.jpg", "media/50a4aeac90bf6bde5f26/F6.jpg", "media/50a4aeac90bf6bde5f26/F7.jpg", "media/50a4aeac90bf6bde5f26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (white scooter with rider in black jacket and white helmet) and object B (red scooter with rider in dark clothing and white helmet) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2306e38775cdf96099b9d2950b3b8167b6548c7dcec10699e60876b535bdb45f"}
{"qa_id": "MOSE__f0b922bf__object_travel_distance_comparison__04", "video_id": "MOSE__f0b922bf", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white scooter with rider in black jacket and white helmet) or B (red scooter with rider in dark clothing and white helmet)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/50a4aeac90bf6bde5f26/F0.jpg", "media/50a4aeac90bf6bde5f26/F1.jpg", "media/50a4aeac90bf6bde5f26/F2.jpg", "media/50a4aeac90bf6bde5f26/F3.jpg", "media/50a4aeac90bf6bde5f26/F4.jpg", "media/50a4aeac90bf6bde5f26/F5.jpg", "media/50a4aeac90bf6bde5f26/F6.jpg", "media/50a4aeac90bf6bde5f26/F7.jpg", "media/50a4aeac90bf6bde5f26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white scooter with rider in black jacket and white helmet) or B (red scooter with rider in dark clothing and white helmet)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0f1557ef60d6301ae635647816785e117ef53b1d939b02c564852aba9410fdb9"}
{"qa_id": "MOSE__f22c39ce__object_world_motion_direction__00", "video_id": "MOSE__f22c39ce", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object B (small elephant walking on the road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "forward", "down", "up"], "frame_paths": ["media/d2e45ce056cf2d8eeedf/F0.jpg", "media/d2e45ce056cf2d8eeedf/F1.jpg", "media/d2e45ce056cf2d8eeedf/F2.jpg", "media/d2e45ce056cf2d8eeedf/F3.jpg", "media/d2e45ce056cf2d8eeedf/F4.jpg", "media/d2e45ce056cf2d8eeedf/F5.jpg", "media/d2e45ce056cf2d8eeedf/F6.jpg", "media/d2e45ce056cf2d8eeedf/F7.jpg", "media/d2e45ce056cf2d8eeedf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object B (small elephant walking on the road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a3d7f1ca6b36157dfaf3fc747d9ef876418788996e62c01a48158409e5b701a9"}
{"qa_id": "MOSE__f2de4198__object_world_motion_direction__00", "video_id": "MOSE__f2de4198", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object B (person in light grey t-shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/537cd5f8c644d7499c32/F0.jpg", "media/537cd5f8c644d7499c32/F1.jpg", "media/537cd5f8c644d7499c32/F2.jpg", "media/537cd5f8c644d7499c32/F3.jpg", "media/537cd5f8c644d7499c32/F4.jpg", "media/537cd5f8c644d7499c32/F5.jpg", "media/537cd5f8c644d7499c32/F6.jpg", "media/537cd5f8c644d7499c32/F7.jpg", "media/537cd5f8c644d7499c32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object B (person in light grey t-shirt) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9404aad780dee7a4824e0b82f29c2c66ff5a91839e8bc7a7245ede37a9405fc9"}
{"qa_id": "MOSE__f2de4198__camera_object_distance_change__01", "video_id": "MOSE__f2de4198", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F6, did object B (person in light grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/537cd5f8c644d7499c32/F0.jpg", "media/537cd5f8c644d7499c32/F1.jpg", "media/537cd5f8c644d7499c32/F2.jpg", "media/537cd5f8c644d7499c32/F3.jpg", "media/537cd5f8c644d7499c32/F4.jpg", "media/537cd5f8c644d7499c32/F5.jpg", "media/537cd5f8c644d7499c32/F6.jpg", "media/537cd5f8c644d7499c32/F7.jpg", "media/537cd5f8c644d7499c32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object B (person in light grey t-shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1317adcd3cb7b5441ee8b450c2f581550900b496e97bcaa34686dacf2a792c81"}
{"qa_id": "MOSE__f2de4198__object_pair_distance_change__02", "video_id": "MOSE__f2de4198", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F6, did the 3D distance between object A (person in white t-shirt and black pants) and object B (person in light grey t-shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/247113848cd3fe4a229e/F0.jpg", "media/247113848cd3fe4a229e/F1.jpg", "media/247113848cd3fe4a229e/F2.jpg", "media/247113848cd3fe4a229e/F3.jpg", "media/247113848cd3fe4a229e/F4.jpg", "media/247113848cd3fe4a229e/F5.jpg", "media/247113848cd3fe4a229e/F6.jpg", "media/247113848cd3fe4a229e/F7.jpg", "media/247113848cd3fe4a229e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did the 3D distance between object A (person in white t-shirt and black pants) and object B (person in light grey t-shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "238da6a803f3b0ac4da9a626507772cbde333db292a71298350d99d24971726e"}
{"qa_id": "MOSE__f2de4198__object_travel_distance_comparison__03", "video_id": "MOSE__f2de4198", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white t-shirt and black pants) or B (person in light grey t-shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/247113848cd3fe4a229e/F0.jpg", "media/247113848cd3fe4a229e/F1.jpg", "media/247113848cd3fe4a229e/F2.jpg", "media/247113848cd3fe4a229e/F3.jpg", "media/247113848cd3fe4a229e/F4.jpg", "media/247113848cd3fe4a229e/F5.jpg", "media/247113848cd3fe4a229e/F6.jpg", "media/247113848cd3fe4a229e/F7.jpg", "media/247113848cd3fe4a229e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white t-shirt and black pants) or B (person in light grey t-shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0413cd78d7a41bfc60394fd31e2088f1f9611c054e6ef0fe99e4ebca9ac4af7"}
{"qa_id": "MOSE__f33d3c15__camera_translation_direction__00", "video_id": "MOSE__f33d3c15", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "left", "up"], "frame_paths": ["media/9cfe53c2fd0461c69c99/F0.jpg", "media/9cfe53c2fd0461c69c99/F1.jpg", "media/9cfe53c2fd0461c69c99/F2.jpg", "media/9cfe53c2fd0461c69c99/F3.jpg", "media/9cfe53c2fd0461c69c99/F4.jpg", "media/9cfe53c2fd0461c69c99/F5.jpg", "media/9cfe53c2fd0461c69c99/F6.jpg", "media/9cfe53c2fd0461c69c99/F7.jpg", "media/9cfe53c2fd0461c69c99/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f27ba5ac870bd56f9f7f9e4013d731ac509f5b0773cc34e3c9e00e9ef5e4f99b"}
{"qa_id": "MOSE__f3978ade__object_pair_distance_change__00", "video_id": "MOSE__f3978ade", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (man in grey t-shirt and jeans sitting on a motorcycle) and object B (man in white t-shirt and dark pants standing with back to camera) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f6f472b41705c06b80fe/F0.jpg", "media/f6f472b41705c06b80fe/F1.jpg", "media/f6f472b41705c06b80fe/F2.jpg", "media/f6f472b41705c06b80fe/F3.jpg", "media/f6f472b41705c06b80fe/F4.jpg", "media/f6f472b41705c06b80fe/F5.jpg", "media/f6f472b41705c06b80fe/F6.jpg", "media/f6f472b41705c06b80fe/F7.jpg", "media/f6f472b41705c06b80fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g5", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (man in grey t-shirt and jeans sitting on a motorcycle) and object B (man in white t-shirt and dark pants standing with back to camera) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "20c9593cfb929095c90b287ec6d96b8ae3f012f112d206802c2454d47986f9a8"}
{"qa_id": "MOSE__f3978ade__object_travel_distance_comparison__01", "video_id": "MOSE__f3978ade", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey t-shirt and jeans sitting on a motorcycle) or B (man in white t-shirt and dark pants standing with back to camera)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/f6f472b41705c06b80fe/F0.jpg", "media/f6f472b41705c06b80fe/F1.jpg", "media/f6f472b41705c06b80fe/F2.jpg", "media/f6f472b41705c06b80fe/F3.jpg", "media/f6f472b41705c06b80fe/F4.jpg", "media/f6f472b41705c06b80fe/F5.jpg", "media/f6f472b41705c06b80fe/F6.jpg", "media/f6f472b41705c06b80fe/F7.jpg", "media/f6f472b41705c06b80fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g5", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in grey t-shirt and jeans sitting on a motorcycle) or B (man in white t-shirt and dark pants standing with back to camera)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31d01877b06da23f4d59fa042ce4404aef83e0b8b9a43f35f252035e00e6b680"}
{"qa_id": "MOSE__f4c4a2ca__camera_translation_direction__00", "video_id": "MOSE__f4c4a2ca", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "forward", "down", "left"], "frame_paths": ["media/b81da475388939b87a60/F0.jpg", "media/b81da475388939b87a60/F1.jpg", "media/b81da475388939b87a60/F2.jpg", "media/b81da475388939b87a60/F3.jpg", "media/b81da475388939b87a60/F4.jpg", "media/b81da475388939b87a60/F5.jpg", "media/b81da475388939b87a60/F6.jpg", "media/b81da475388939b87a60/F7.jpg", "media/b81da475388939b87a60/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5aa1c8035c8044e26c75f478ee262c6c6516678474083a7a976cdc9e92dbdfb"}
{"qa_id": "MOSE__f521e3c5__camera_translation_direction__00", "video_id": "MOSE__f521e3c5", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "down", "right", "up"], "frame_paths": ["media/fe98eaf4314f6045bf52/F0.jpg", "media/fe98eaf4314f6045bf52/F1.jpg", "media/fe98eaf4314f6045bf52/F2.jpg", "media/fe98eaf4314f6045bf52/F3.jpg", "media/fe98eaf4314f6045bf52/F4.jpg", "media/fe98eaf4314f6045bf52/F5.jpg", "media/fe98eaf4314f6045bf52/F6.jpg", "media/fe98eaf4314f6045bf52/F7.jpg", "media/fe98eaf4314f6045bf52/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "502fabef7540724a5dff9809b4836cf7b4e425f4a06400c5bc3540aad8223374"}
{"qa_id": "MOSE__f521e3c5__camera_object_distance_change__01", "video_id": "MOSE__f521e3c5", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object A (man in grey plaid suit) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/edb85f7daa20ea53fd98/F0.jpg", "media/edb85f7daa20ea53fd98/F1.jpg", "media/edb85f7daa20ea53fd98/F2.jpg", "media/edb85f7daa20ea53fd98/F3.jpg", "media/edb85f7daa20ea53fd98/F4.jpg", "media/edb85f7daa20ea53fd98/F5.jpg", "media/edb85f7daa20ea53fd98/F6.jpg", "media/edb85f7daa20ea53fd98/F7.jpg", "media/edb85f7daa20ea53fd98/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (man in grey plaid suit) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "500066b5ecf45308014a142fee8dd9c815d366f753d904d973d65b733d4ce9d4"}
{"qa_id": "MOSE__f521e3c5__camera_object_distance_change__02", "video_id": "MOSE__f521e3c5", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object B (man in maroon traditional robe) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b8d91ca06458216518a9/F0.jpg", "media/b8d91ca06458216518a9/F1.jpg", "media/b8d91ca06458216518a9/F2.jpg", "media/b8d91ca06458216518a9/F3.jpg", "media/b8d91ca06458216518a9/F4.jpg", "media/b8d91ca06458216518a9/F5.jpg", "media/b8d91ca06458216518a9/F6.jpg", "media/b8d91ca06458216518a9/F7.jpg", "media/b8d91ca06458216518a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (man in maroon traditional robe) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eef063ffa58dc1650782e40b44cf8e357a958f85859368fd1b5018faa371062f"}
{"qa_id": "MOSE__f56942e9__camera_translation_direction__00", "video_id": "MOSE__f56942e9", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "forward", "up", "left"], "frame_paths": ["media/357b3d8d6d3129026f2f/F0.jpg", "media/357b3d8d6d3129026f2f/F1.jpg", "media/357b3d8d6d3129026f2f/F2.jpg", "media/357b3d8d6d3129026f2f/F3.jpg", "media/357b3d8d6d3129026f2f/F4.jpg", "media/357b3d8d6d3129026f2f/F5.jpg", "media/357b3d8d6d3129026f2f/F6.jpg", "media/357b3d8d6d3129026f2f/F7.jpg", "media/357b3d8d6d3129026f2f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "de6f05c1390c06bb8a33201525187410fbb5347e5b8b4f4992f30fe4598442f2"}
{"qa_id": "MOSE__f6786ab5__object_world_motion_direction__01", "video_id": "MOSE__f6786ab5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object B (large adult elephant walking on the dirt shoulder) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "down", "left", "forward"], "frame_paths": ["media/e9a140dfae578169520e/F0.jpg", "media/e9a140dfae578169520e/F1.jpg", "media/e9a140dfae578169520e/F2.jpg", "media/e9a140dfae578169520e/F3.jpg", "media/e9a140dfae578169520e/F4.jpg", "media/e9a140dfae578169520e/F5.jpg", "media/e9a140dfae578169520e/F6.jpg", "media/e9a140dfae578169520e/F7.jpg", "media/e9a140dfae578169520e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object B (large adult elephant walking on the dirt shoulder) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b6f10a4f447397c874718dcd57c0b4fe7119966a423c4ae97732a91a7b63f91e"}
{"qa_id": "MOSE__f6e40aba__object_pair_distance_change__00", "video_id": "MOSE__f6e40aba", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F6, did the 3D distance between object A (person with dark hair wearing a black long-sleeved top, crouching behind a wire fence) and object B (brown chicken standing near a white feeding bowl) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/8ba79e9c432873056d62/F0.jpg", "media/8ba79e9c432873056d62/F1.jpg", "media/8ba79e9c432873056d62/F2.jpg", "media/8ba79e9c432873056d62/F3.jpg", "media/8ba79e9c432873056d62/F4.jpg", "media/8ba79e9c432873056d62/F5.jpg", "media/8ba79e9c432873056d62/F6.jpg", "media/8ba79e9c432873056d62/F7.jpg", "media/8ba79e9c432873056d62/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did the 3D distance between object A (person with dark hair wearing a black long-sleeved top, crouching behind a wire fence) and object B (brown chicken standing near a white feeding bowl) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "512a3f5bac790a6c6261c6cc765a94074d6dd459f7178a17243e727b2ee97cee"}
{"qa_id": "MOSE__f7360199__object_world_motion_direction__00", "video_id": "MOSE__f7360199", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (white China Eastern airplane) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "left", "forward", "up", "backward", "right"], "frame_paths": ["media/08a1539404c12cc97e15/F0.jpg", "media/08a1539404c12cc97e15/F1.jpg", "media/08a1539404c12cc97e15/F2.jpg", "media/08a1539404c12cc97e15/F3.jpg", "media/08a1539404c12cc97e15/F4.jpg", "media/08a1539404c12cc97e15/F5.jpg", "media/08a1539404c12cc97e15/F6.jpg", "media/08a1539404c12cc97e15/F7.jpg", "media/08a1539404c12cc97e15/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (white China Eastern airplane) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. forward\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "67566f6d6b10067c271e9eeb22ba782dc1385c01e128e5cf42b2168510f8db96"}
{"qa_id": "MOSE__f7935d24__camera_translation_direction__00", "video_id": "MOSE__f7935d24", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "backward", "forward"], "frame_paths": ["media/7c1741f32c693863487e/F0.jpg", "media/7c1741f32c693863487e/F1.jpg", "media/7c1741f32c693863487e/F2.jpg", "media/7c1741f32c693863487e/F3.jpg", "media/7c1741f32c693863487e/F4.jpg", "media/7c1741f32c693863487e/F5.jpg", "media/7c1741f32c693863487e/F6.jpg", "media/7c1741f32c693863487e/F7.jpg", "media/7c1741f32c693863487e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2f4ba3f9ff062d56ad5a27aa85e5f9d647257d8ddd6aa66069a4d0e23d98d1f5"}
{"qa_id": "MOSE__f7935d24__camera_object_distance_change__01", "video_id": "MOSE__f7935d24", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (human arm reaching towards the cats) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d182825139ea7b779011/F0.jpg", "media/d182825139ea7b779011/F1.jpg", "media/d182825139ea7b779011/F2.jpg", "media/d182825139ea7b779011/F3.jpg", "media/d182825139ea7b779011/F4.jpg", "media/d182825139ea7b779011/F5.jpg", "media/d182825139ea7b779011/F6.jpg", "media/d182825139ea7b779011/F7.jpg", "media/d182825139ea7b779011/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (human arm reaching towards the cats) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "24b9e682787dfac2d874e66e35bc398ceedf26c1f667f2c05c61764df49250f2"}
{"qa_id": "MOSE__f7935d24__object_pair_distance_change__02", "video_id": "MOSE__f7935d24", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did the 3D distance between object A (human arm reaching towards the cats) and object B (tabby cat lying on the wooden table) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/126e6d9e24dadfec5b4f/F0.jpg", "media/126e6d9e24dadfec5b4f/F1.jpg", "media/126e6d9e24dadfec5b4f/F2.jpg", "media/126e6d9e24dadfec5b4f/F3.jpg", "media/126e6d9e24dadfec5b4f/F4.jpg", "media/126e6d9e24dadfec5b4f/F5.jpg", "media/126e6d9e24dadfec5b4f/F6.jpg", "media/126e6d9e24dadfec5b4f/F7.jpg", "media/126e6d9e24dadfec5b4f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (human arm reaching towards the cats) and object B (tabby cat lying on the wooden table) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e83510fc23469774b7416b80fa6b592524c315f6b3932ad1f039d2635e43088a"}
{"qa_id": "MOSE__f7935d24__object_travel_distance_comparison__03", "video_id": "MOSE__f7935d24", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (human arm reaching towards the cats) or B (tabby cat lying on the wooden table)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/126e6d9e24dadfec5b4f/F0.jpg", "media/126e6d9e24dadfec5b4f/F1.jpg", "media/126e6d9e24dadfec5b4f/F2.jpg", "media/126e6d9e24dadfec5b4f/F3.jpg", "media/126e6d9e24dadfec5b4f/F4.jpg", "media/126e6d9e24dadfec5b4f/F5.jpg", "media/126e6d9e24dadfec5b4f/F6.jpg", "media/126e6d9e24dadfec5b4f/F7.jpg", "media/126e6d9e24dadfec5b4f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (human arm reaching towards the cats) or B (tabby cat lying on the wooden table)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "272810c974edc0c3fa715aaba9b4df38a9647283d3a72f66324dd2ae354e213e"}
{"qa_id": "MOSE__f7d36a47__camera_object_distance_change__00", "video_id": "MOSE__f7d36a47", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object A (white minivan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/02b404c1247bab8aa0c5/F0.jpg", "media/02b404c1247bab8aa0c5/F1.jpg", "media/02b404c1247bab8aa0c5/F2.jpg", "media/02b404c1247bab8aa0c5/F3.jpg", "media/02b404c1247bab8aa0c5/F4.jpg", "media/02b404c1247bab8aa0c5/F5.jpg", "media/02b404c1247bab8aa0c5/F6.jpg", "media/02b404c1247bab8aa0c5/F7.jpg", "media/02b404c1247bab8aa0c5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (white minivan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5142032f6d8ee04ecab11f6c6da4ec2e855569297d5efc0b6a5e737dd0a762f2"}
{"qa_id": "MOSE__f7d36a47__object_world_motion_direction__01", "video_id": "MOSE__f7d36a47", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (black hatchback with side decals) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "left", "down", "backward"], "frame_paths": ["media/28da94c63b02da4a1e5b/F0.jpg", "media/28da94c63b02da4a1e5b/F1.jpg", "media/28da94c63b02da4a1e5b/F2.jpg", "media/28da94c63b02da4a1e5b/F3.jpg", "media/28da94c63b02da4a1e5b/F4.jpg", "media/28da94c63b02da4a1e5b/F5.jpg", "media/28da94c63b02da4a1e5b/F6.jpg", "media/28da94c63b02da4a1e5b/F7.jpg", "media/28da94c63b02da4a1e5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (black hatchback with side decals) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a965d172a256e95524f664b91de42987846ae185daa44fd592e9636934710755"}
{"qa_id": "MOSE__f7d36a47__camera_object_distance_change__02", "video_id": "MOSE__f7d36a47", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (black hatchback with side decals) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/28da94c63b02da4a1e5b/F0.jpg", "media/28da94c63b02da4a1e5b/F1.jpg", "media/28da94c63b02da4a1e5b/F2.jpg", "media/28da94c63b02da4a1e5b/F3.jpg", "media/28da94c63b02da4a1e5b/F4.jpg", "media/28da94c63b02da4a1e5b/F5.jpg", "media/28da94c63b02da4a1e5b/F6.jpg", "media/28da94c63b02da4a1e5b/F7.jpg", "media/28da94c63b02da4a1e5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (black hatchback with side decals) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a7ae53e7c28a21efe4a0a1d1bd4e79dc5c87e72d0efff742a7395953a0cf29e9"}
{"qa_id": "MOSE__f7d36a47__object_pair_distance_change__03", "video_id": "MOSE__f7d36a47", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (white minivan) and object B (black hatchback with side decals) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/56a85bc0fcd86815c95e/F0.jpg", "media/56a85bc0fcd86815c95e/F1.jpg", "media/56a85bc0fcd86815c95e/F2.jpg", "media/56a85bc0fcd86815c95e/F3.jpg", "media/56a85bc0fcd86815c95e/F4.jpg", "media/56a85bc0fcd86815c95e/F5.jpg", "media/56a85bc0fcd86815c95e/F6.jpg", "media/56a85bc0fcd86815c95e/F7.jpg", "media/56a85bc0fcd86815c95e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (white minivan) and object B (black hatchback with side decals) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "69c0e8208d341b0258e9dce84d82c9e79392c3ea95584895b84eae3348c80ba0"}
{"qa_id": "MOSE__f8916fef__camera_translation_direction__00", "video_id": "MOSE__f8916fef", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "down", "backward"], "frame_paths": ["media/ef8efd1625728911a409/F0.jpg", "media/ef8efd1625728911a409/F1.jpg", "media/ef8efd1625728911a409/F2.jpg", "media/ef8efd1625728911a409/F3.jpg", "media/ef8efd1625728911a409/F4.jpg", "media/ef8efd1625728911a409/F5.jpg", "media/ef8efd1625728911a409/F6.jpg", "media/ef8efd1625728911a409/F7.jpg", "media/ef8efd1625728911a409/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3c1ded3f6ca30dbff7563e5e0ed096297e5f3dfb852f52be44e8561968cc2fd4"}
{"qa_id": "MOSE__f8916fef__camera_view_direction_change__01", "video_id": "MOSE__f8916fef", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/ef8efd1625728911a409/F0.jpg", "media/ef8efd1625728911a409/F1.jpg", "media/ef8efd1625728911a409/F2.jpg", "media/ef8efd1625728911a409/F3.jpg", "media/ef8efd1625728911a409/F4.jpg", "media/ef8efd1625728911a409/F5.jpg", "media/ef8efd1625728911a409/F6.jpg", "media/ef8efd1625728911a409/F7.jpg", "media/ef8efd1625728911a409/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4bb7ec21e095c5c46cce0347f69602ee73fee0909fed7189c01b64cbf085997"}
{"qa_id": "MOSE__f8916fef__camera_object_distance_change__02", "video_id": "MOSE__f8916fef", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did object B (woman in a blue denim dress with white sleeves) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f84cac6aa5e86f5f72f6/F0.jpg", "media/f84cac6aa5e86f5f72f6/F1.jpg", "media/f84cac6aa5e86f5f72f6/F2.jpg", "media/f84cac6aa5e86f5f72f6/F3.jpg", "media/f84cac6aa5e86f5f72f6/F4.jpg", "media/f84cac6aa5e86f5f72f6/F5.jpg", "media/f84cac6aa5e86f5f72f6/F6.jpg", "media/f84cac6aa5e86f5f72f6/F7.jpg", "media/f84cac6aa5e86f5f72f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object B (woman in a blue denim dress with white sleeves) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "693b901b34803a5e3684ac9dcfc0dd40bb9f115ed1caf0d996997ffeb0b7549d"}
{"qa_id": "MOSE__f8916fef__object_pair_distance_change__03", "video_id": "MOSE__f8916fef", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did the 3D distance between object A (man in a dark plaid suit jacket and glasses) and object B (woman in a blue denim dress with white sleeves) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/620c08eaaa42cbed4b47/F0.jpg", "media/620c08eaaa42cbed4b47/F1.jpg", "media/620c08eaaa42cbed4b47/F2.jpg", "media/620c08eaaa42cbed4b47/F3.jpg", "media/620c08eaaa42cbed4b47/F4.jpg", "media/620c08eaaa42cbed4b47/F5.jpg", "media/620c08eaaa42cbed4b47/F6.jpg", "media/620c08eaaa42cbed4b47/F7.jpg", "media/620c08eaaa42cbed4b47/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (man in a dark plaid suit jacket and glasses) and object B (woman in a blue denim dress with white sleeves) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78a15b52c5e1beb1a650d9ae64888a4136ccecf73392c3b8a763c41736d90864"}
{"qa_id": "MOSE__f8916fef__object_travel_distance_comparison__04", "video_id": "MOSE__f8916fef", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in a dark plaid suit jacket and glasses) or B (woman in a blue denim dress with white sleeves)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/620c08eaaa42cbed4b47/F0.jpg", "media/620c08eaaa42cbed4b47/F1.jpg", "media/620c08eaaa42cbed4b47/F2.jpg", "media/620c08eaaa42cbed4b47/F3.jpg", "media/620c08eaaa42cbed4b47/F4.jpg", "media/620c08eaaa42cbed4b47/F5.jpg", "media/620c08eaaa42cbed4b47/F6.jpg", "media/620c08eaaa42cbed4b47/F7.jpg", "media/620c08eaaa42cbed4b47/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in a dark plaid suit jacket and glasses) or B (woman in a blue denim dress with white sleeves)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "70695688b6bef552773d0fd40745f71caf4b7ef06d291200d8d38ee0b97972dd"}
{"qa_id": "MOSE__f8e5f837__camera_translation_direction__00", "video_id": "MOSE__f8e5f837", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "down", "forward", "right"], "frame_paths": ["media/8994197d242c0f7e76fe/F0.jpg", "media/8994197d242c0f7e76fe/F1.jpg", "media/8994197d242c0f7e76fe/F2.jpg", "media/8994197d242c0f7e76fe/F3.jpg", "media/8994197d242c0f7e76fe/F4.jpg", "media/8994197d242c0f7e76fe/F5.jpg", "media/8994197d242c0f7e76fe/F6.jpg", "media/8994197d242c0f7e76fe/F7.jpg", "media/8994197d242c0f7e76fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b760a72010dfa91910ad8aece896ddf005514fc0c580d314ee3f69b9c463d5d"}
{"qa_id": "MOSE__fa95a6a2__camera_translation_direction__00", "video_id": "MOSE__fa95a6a2", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "left", "up", "forward"], "frame_paths": ["media/6d5fd3365884ba23e9f5/F0.jpg", "media/6d5fd3365884ba23e9f5/F1.jpg", "media/6d5fd3365884ba23e9f5/F2.jpg", "media/6d5fd3365884ba23e9f5/F3.jpg", "media/6d5fd3365884ba23e9f5/F4.jpg", "media/6d5fd3365884ba23e9f5/F5.jpg", "media/6d5fd3365884ba23e9f5/F6.jpg", "media/6d5fd3365884ba23e9f5/F7.jpg", "media/6d5fd3365884ba23e9f5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d162dcebaa238fc474e78c186fac9ba3e02785875d77efb00e6f3bb59c84d9dd"}
{"qa_id": "MOSE__fabae187__object_world_motion_direction__00", "video_id": "MOSE__fabae187", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object B (large red fish on the right side) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "down", "forward"], "frame_paths": ["media/d0a4848a97f1456beffe/F0.jpg", "media/d0a4848a97f1456beffe/F1.jpg", "media/d0a4848a97f1456beffe/F2.jpg", "media/d0a4848a97f1456beffe/F3.jpg", "media/d0a4848a97f1456beffe/F4.jpg", "media/d0a4848a97f1456beffe/F5.jpg", "media/d0a4848a97f1456beffe/F6.jpg", "media/d0a4848a97f1456beffe/F7.jpg", "media/d0a4848a97f1456beffe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g9"}, "requested_object_ids": ["g9"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object B (large red fish on the right side) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2c35243ddb0d7c4ba8e7444f4b10c26c69048e33273443d2f1faad1b03675f53"}
{"qa_id": "MOSE__fabae187__object_travel_distance_comparison__01", "video_id": "MOSE__fabae187", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small red fish at the bottom left) or B (large red fish on the right side)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/6e9eace76f7572edb4e9/F0.jpg", "media/6e9eace76f7572edb4e9/F1.jpg", "media/6e9eace76f7572edb4e9/F2.jpg", "media/6e9eace76f7572edb4e9/F3.jpg", "media/6e9eace76f7572edb4e9/F4.jpg", "media/6e9eace76f7572edb4e9/F5.jpg", "media/6e9eace76f7572edb4e9/F6.jpg", "media/6e9eace76f7572edb4e9/F7.jpg", "media/6e9eace76f7572edb4e9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g9"}, "requested_object_ids": ["g3", "g9"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small red fish at the bottom left) or B (large red fish on the right side)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bbbbfa593e5388ca8f8441215ba3c7a1a8d79df14b7c6bd2c3cd1030cbe25d4d"}
{"qa_id": "MOSE__faf7d852__camera_translation_direction__00", "video_id": "MOSE__faf7d852", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "down", "up"], "frame_paths": ["media/55fca133d6bcf1eb0fa4/F0.jpg", "media/55fca133d6bcf1eb0fa4/F1.jpg", "media/55fca133d6bcf1eb0fa4/F2.jpg", "media/55fca133d6bcf1eb0fa4/F3.jpg", "media/55fca133d6bcf1eb0fa4/F4.jpg", "media/55fca133d6bcf1eb0fa4/F5.jpg", "media/55fca133d6bcf1eb0fa4/F6.jpg", "media/55fca133d6bcf1eb0fa4/F7.jpg", "media/55fca133d6bcf1eb0fa4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1662dafc78aa30c1ab3db648a6de1be6a451455f53cbeb60ce8920c7b3921d8"}
{"qa_id": "MOSE__fb16e746__camera_translation_direction__00", "video_id": "MOSE__fb16e746", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/ebc72ecef991ad642289/F0.jpg", "media/ebc72ecef991ad642289/F1.jpg", "media/ebc72ecef991ad642289/F2.jpg", "media/ebc72ecef991ad642289/F3.jpg", "media/ebc72ecef991ad642289/F4.jpg", "media/ebc72ecef991ad642289/F5.jpg", "media/ebc72ecef991ad642289/F6.jpg", "media/ebc72ecef991ad642289/F7.jpg", "media/ebc72ecef991ad642289/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "58ead3af71162682c16700df846e1634b0c87aa93b7423ce20eaae292910fa9a"}
{"qa_id": "MOSE__fbfb6e30__object_pair_distance_change__00", "video_id": "MOSE__fbfb6e30", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (zebra on the right side of the group, head turned towards the camera) and object B (zebra in the center-left, standing behind the fence with head visible) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/40bcac36326b5c1303f0/F0.jpg", "media/40bcac36326b5c1303f0/F1.jpg", "media/40bcac36326b5c1303f0/F2.jpg", "media/40bcac36326b5c1303f0/F3.jpg", "media/40bcac36326b5c1303f0/F4.jpg", "media/40bcac36326b5c1303f0/F5.jpg", "media/40bcac36326b5c1303f0/F6.jpg", "media/40bcac36326b5c1303f0/F7.jpg", "media/40bcac36326b5c1303f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (zebra on the right side of the group, head turned towards the camera) and object B (zebra in the center-left, standing behind the fence with head visible) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da421789343a2e907f9725d79ce3819dd9363dbc14189706ff9aab1d805be982"}
{"qa_id": "MOSE__fc4c1c67__camera_object_distance_change__00", "video_id": "MOSE__fc4c1c67", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (tiger walking towards the camera) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/edb5427c6c49ed0524d6/F0.jpg", "media/edb5427c6c49ed0524d6/F1.jpg", "media/edb5427c6c49ed0524d6/F2.jpg", "media/edb5427c6c49ed0524d6/F3.jpg", "media/edb5427c6c49ed0524d6/F4.jpg", "media/edb5427c6c49ed0524d6/F5.jpg", "media/edb5427c6c49ed0524d6/F6.jpg", "media/edb5427c6c49ed0524d6/F7.jpg", "media/edb5427c6c49ed0524d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (tiger walking towards the camera) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee2b19a84b3f005858dcea343af85ec9d4bb73ab73f84d16931e2f2c137d1aeb"}
{"qa_id": "MOSE__fc4c1c67__object_travel_distance_comparison__01", "video_id": "MOSE__fc4c1c67", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tiger walking towards the camera) or B (tiger standing on the left side)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/2a6246baa3d398371e04/F0.jpg", "media/2a6246baa3d398371e04/F1.jpg", "media/2a6246baa3d398371e04/F2.jpg", "media/2a6246baa3d398371e04/F3.jpg", "media/2a6246baa3d398371e04/F4.jpg", "media/2a6246baa3d398371e04/F5.jpg", "media/2a6246baa3d398371e04/F6.jpg", "media/2a6246baa3d398371e04/F7.jpg", "media/2a6246baa3d398371e04/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tiger walking towards the camera) or B (tiger standing on the left side)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "28ede633e29018a1efbb0b718eebd6f23c62cf9297cab90c170f77e746ae3748"}
{"qa_id": "MOSE__fc79c03e__camera_view_direction_change__00", "video_id": "MOSE__fc79c03e", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/e3ae7f0c37058b062c7d/F0.jpg", "media/e3ae7f0c37058b062c7d/F1.jpg", "media/e3ae7f0c37058b062c7d/F2.jpg", "media/e3ae7f0c37058b062c7d/F3.jpg", "media/e3ae7f0c37058b062c7d/F4.jpg", "media/e3ae7f0c37058b062c7d/F5.jpg", "media/e3ae7f0c37058b062c7d/F6.jpg", "media/e3ae7f0c37058b062c7d/F7.jpg", "media/e3ae7f0c37058b062c7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9557c0796e28ae2a4f32990ec2b747f1a5933b7979d70ae5cf47b924054e1483"}
{"qa_id": "MOSE__fe2dc281__object_world_motion_direction__00", "video_id": "MOSE__fe2dc281", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (small white box truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "left", "up", "forward"], "frame_paths": ["media/30b5d64af8e0fdbcc8d6/F0.jpg", "media/30b5d64af8e0fdbcc8d6/F1.jpg", "media/30b5d64af8e0fdbcc8d6/F2.jpg", "media/30b5d64af8e0fdbcc8d6/F3.jpg", "media/30b5d64af8e0fdbcc8d6/F4.jpg", "media/30b5d64af8e0fdbcc8d6/F5.jpg", "media/30b5d64af8e0fdbcc8d6/F6.jpg", "media/30b5d64af8e0fdbcc8d6/F7.jpg", "media/30b5d64af8e0fdbcc8d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (small white box truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f30f2bd105e73f6fe460f746ad5ae04d5d315af04f597c94aa1a3a224c3268d"}
{"qa_id": "MOSE__fe2dc281__camera_object_distance_change__01", "video_id": "MOSE__fe2dc281", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object A (small white box truck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/30b5d64af8e0fdbcc8d6/F0.jpg", "media/30b5d64af8e0fdbcc8d6/F1.jpg", "media/30b5d64af8e0fdbcc8d6/F2.jpg", "media/30b5d64af8e0fdbcc8d6/F3.jpg", "media/30b5d64af8e0fdbcc8d6/F4.jpg", "media/30b5d64af8e0fdbcc8d6/F5.jpg", "media/30b5d64af8e0fdbcc8d6/F6.jpg", "media/30b5d64af8e0fdbcc8d6/F7.jpg", "media/30b5d64af8e0fdbcc8d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (small white box truck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6410464424e682b007cb9c6399af569953dd51442bf5ad23c9b7a0f48df90865"}
{"qa_id": "MOSE__fe470104__object_pair_distance_change__00", "video_id": "MOSE__fe470104", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F7, did the 3D distance between object A (giraffe standing on the far right) and object B (giraffe standing in the left-center foreground) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/4c2c60b87ff4d065ce88/F0.jpg", "media/4c2c60b87ff4d065ce88/F1.jpg", "media/4c2c60b87ff4d065ce88/F2.jpg", "media/4c2c60b87ff4d065ce88/F3.jpg", "media/4c2c60b87ff4d065ce88/F4.jpg", "media/4c2c60b87ff4d065ce88/F5.jpg", "media/4c2c60b87ff4d065ce88/F6.jpg", "media/4c2c60b87ff4d065ce88/F7.jpg", "media/4c2c60b87ff4d065ce88/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did the 3D distance between object A (giraffe standing on the far right) and object B (giraffe standing in the left-center foreground) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eee74fe041d9a3256c613931dd359406dc27720d60df1f58437b8127b7703822"}
{"qa_id": "MOSE__ff041cd1__camera_view_direction_change__00", "video_id": "MOSE__ff041cd1", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/06fb70a9ffb852a14943/F0.jpg", "media/06fb70a9ffb852a14943/F1.jpg", "media/06fb70a9ffb852a14943/F2.jpg", "media/06fb70a9ffb852a14943/F3.jpg", "media/06fb70a9ffb852a14943/F4.jpg", "media/06fb70a9ffb852a14943/F5.jpg", "media/06fb70a9ffb852a14943/F6.jpg", "media/06fb70a9ffb852a14943/F7.jpg", "media/06fb70a9ffb852a14943/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1651fff05924a19e0836ca818b253e1508fc709067ed9c52ff7f7e8d9f60bdb"}
{"qa_id": "MOSE__ff1142df__object_world_motion_direction__00", "video_id": "MOSE__ff1142df", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (black bear walking on the grass) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "right", "left", "up"], "frame_paths": ["media/dcb5deea5078b1e10e07/F0.jpg", "media/dcb5deea5078b1e10e07/F1.jpg", "media/dcb5deea5078b1e10e07/F2.jpg", "media/dcb5deea5078b1e10e07/F3.jpg", "media/dcb5deea5078b1e10e07/F4.jpg", "media/dcb5deea5078b1e10e07/F5.jpg", "media/dcb5deea5078b1e10e07/F6.jpg", "media/dcb5deea5078b1e10e07/F7.jpg", "media/dcb5deea5078b1e10e07/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (black bear walking on the grass) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a1a63cecb293dfd7e890f352de008919346e30fcd9004f5ca70876180f2b17a"}
{"qa_id": "MOSE__ff1142df__object_pair_distance_change__01", "video_id": "MOSE__ff1142df", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (black bear walking on the grass) and object B (black bear near the rocks on the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2ba4067a8966423a06c1/F0.jpg", "media/2ba4067a8966423a06c1/F1.jpg", "media/2ba4067a8966423a06c1/F2.jpg", "media/2ba4067a8966423a06c1/F3.jpg", "media/2ba4067a8966423a06c1/F4.jpg", "media/2ba4067a8966423a06c1/F5.jpg", "media/2ba4067a8966423a06c1/F6.jpg", "media/2ba4067a8966423a06c1/F7.jpg", "media/2ba4067a8966423a06c1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (black bear walking on the grass) and object B (black bear near the rocks on the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ad97bd055bc73e3dbb51ecc4bc175cdcd39f9ca3c030d8e31961b3c6ff9c6009"}
{"qa_id": "MOSE__ff1142df__object_travel_distance_comparison__02", "video_id": "MOSE__ff1142df", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black bear walking on the grass) or B (black bear near the rocks on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/2ba4067a8966423a06c1/F0.jpg", "media/2ba4067a8966423a06c1/F1.jpg", "media/2ba4067a8966423a06c1/F2.jpg", "media/2ba4067a8966423a06c1/F3.jpg", "media/2ba4067a8966423a06c1/F4.jpg", "media/2ba4067a8966423a06c1/F5.jpg", "media/2ba4067a8966423a06c1/F6.jpg", "media/2ba4067a8966423a06c1/F7.jpg", "media/2ba4067a8966423a06c1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black bear walking on the grass) or B (black bear near the rocks on the left)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cae21162bb320c543fb6448e74dce93c95f3c252f67e88cbad606cdde5d1f4d3"}
{"qa_id": "MVS-Synth__0000__camera_view_direction_change__00", "video_id": "MVS-Synth__0000", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/2a3ecf8a454c782875f2/F0.jpg", "media/2a3ecf8a454c782875f2/F1.jpg", "media/2a3ecf8a454c782875f2/F2.jpg", "media/2a3ecf8a454c782875f2/F3.jpg", "media/2a3ecf8a454c782875f2/F4.jpg", "media/2a3ecf8a454c782875f2/F5.jpg", "media/2a3ecf8a454c782875f2/F6.jpg", "media/2a3ecf8a454c782875f2/F7.jpg", "media/2a3ecf8a454c782875f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "feac7bb2bfacc4a82c8f095b3582ce98d13b53eeba85dec66376786f8cb65725"}
{"qa_id": "MVS-Synth__0001__camera_translation_direction__00", "video_id": "MVS-Synth__0001", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "forward", "backward"], "frame_paths": ["media/89eee1ad0d9b269a8373/F0.jpg", "media/89eee1ad0d9b269a8373/F1.jpg", "media/89eee1ad0d9b269a8373/F2.jpg", "media/89eee1ad0d9b269a8373/F3.jpg", "media/89eee1ad0d9b269a8373/F4.jpg", "media/89eee1ad0d9b269a8373/F5.jpg", "media/89eee1ad0d9b269a8373/F6.jpg", "media/89eee1ad0d9b269a8373/F7.jpg", "media/89eee1ad0d9b269a8373/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "99da4d4ab24ad8292052a93ca362bd1a1ada4c2acc5c9baed7a1f0e14173bc9c"}
{"qa_id": "MVS-Synth__0001__camera_view_direction_change__01", "video_id": "MVS-Synth__0001", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/89eee1ad0d9b269a8373/F0.jpg", "media/89eee1ad0d9b269a8373/F1.jpg", "media/89eee1ad0d9b269a8373/F2.jpg", "media/89eee1ad0d9b269a8373/F3.jpg", "media/89eee1ad0d9b269a8373/F4.jpg", "media/89eee1ad0d9b269a8373/F5.jpg", "media/89eee1ad0d9b269a8373/F6.jpg", "media/89eee1ad0d9b269a8373/F7.jpg", "media/89eee1ad0d9b269a8373/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5a271205c893da88805ae3ae4c0687f454dd76d1e65adbc019bf91b0eea5fb74"}
{"qa_id": "MVS-Synth__0003__camera_translation_direction__00", "video_id": "MVS-Synth__0003", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/8580dd00c026e645cc2a/F0.jpg", "media/8580dd00c026e645cc2a/F1.jpg", "media/8580dd00c026e645cc2a/F2.jpg", "media/8580dd00c026e645cc2a/F3.jpg", "media/8580dd00c026e645cc2a/F4.jpg", "media/8580dd00c026e645cc2a/F5.jpg", "media/8580dd00c026e645cc2a/F6.jpg", "media/8580dd00c026e645cc2a/F7.jpg", "media/8580dd00c026e645cc2a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6effff33e12e5c996ecebe2687027e9d7d419835815a240bbbbba88e871b24e6"}
{"qa_id": "MVS-Synth__0003__camera_view_direction_change__01", "video_id": "MVS-Synth__0003", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/8580dd00c026e645cc2a/F0.jpg", "media/8580dd00c026e645cc2a/F1.jpg", "media/8580dd00c026e645cc2a/F2.jpg", "media/8580dd00c026e645cc2a/F3.jpg", "media/8580dd00c026e645cc2a/F4.jpg", "media/8580dd00c026e645cc2a/F5.jpg", "media/8580dd00c026e645cc2a/F6.jpg", "media/8580dd00c026e645cc2a/F7.jpg", "media/8580dd00c026e645cc2a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88b40c1ef52643cefe4ef1c3fa249f1a470c7fcca2783a307d04d9b65d9496e2"}
{"qa_id": "MVS-Synth__0004__camera_view_direction_change__00", "video_id": "MVS-Synth__0004", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/ac060e00cc9e1cd845bf/F0.jpg", "media/ac060e00cc9e1cd845bf/F1.jpg", "media/ac060e00cc9e1cd845bf/F2.jpg", "media/ac060e00cc9e1cd845bf/F3.jpg", "media/ac060e00cc9e1cd845bf/F4.jpg", "media/ac060e00cc9e1cd845bf/F5.jpg", "media/ac060e00cc9e1cd845bf/F6.jpg", "media/ac060e00cc9e1cd845bf/F7.jpg", "media/ac060e00cc9e1cd845bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8577f119e60dfe3c6c7b6dbb624c94831b16ba75bf4b87861a80070d2d7143cf"}
{"qa_id": "MVS-Synth__0005__camera_view_direction_change__00", "video_id": "MVS-Synth__0005", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/774534c3c2980534d33e/F0.jpg", "media/774534c3c2980534d33e/F1.jpg", "media/774534c3c2980534d33e/F2.jpg", "media/774534c3c2980534d33e/F3.jpg", "media/774534c3c2980534d33e/F4.jpg", "media/774534c3c2980534d33e/F5.jpg", "media/774534c3c2980534d33e/F6.jpg", "media/774534c3c2980534d33e/F7.jpg", "media/774534c3c2980534d33e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "892b23147d17a5e53e289278ee769dfa3e247cd14585fe2cb2e4458cf11b33c7"}
{"qa_id": "MVS-Synth__0006__camera_translation_direction__00", "video_id": "MVS-Synth__0006", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/e01319f51a322d332265/F0.jpg", "media/e01319f51a322d332265/F1.jpg", "media/e01319f51a322d332265/F2.jpg", "media/e01319f51a322d332265/F3.jpg", "media/e01319f51a322d332265/F4.jpg", "media/e01319f51a322d332265/F5.jpg", "media/e01319f51a322d332265/F6.jpg", "media/e01319f51a322d332265/F7.jpg", "media/e01319f51a322d332265/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0217186dd94e51e0778b0c4cd8bc97948af4b30da77b71e5ea0a476616517991"}
{"qa_id": "MVS-Synth__0006__camera_view_direction_change__01", "video_id": "MVS-Synth__0006", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/e01319f51a322d332265/F0.jpg", "media/e01319f51a322d332265/F1.jpg", "media/e01319f51a322d332265/F2.jpg", "media/e01319f51a322d332265/F3.jpg", "media/e01319f51a322d332265/F4.jpg", "media/e01319f51a322d332265/F5.jpg", "media/e01319f51a322d332265/F6.jpg", "media/e01319f51a322d332265/F7.jpg", "media/e01319f51a322d332265/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7809c30ae0f1d21b4e882cd9ee222f30cccc0e1b8468e72a78dae4e48d0ee4e"}
{"qa_id": "MVS-Synth__0007__camera_view_direction_change__00", "video_id": "MVS-Synth__0007", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/9c32ca83538fa9a793aa/F0.jpg", "media/9c32ca83538fa9a793aa/F1.jpg", "media/9c32ca83538fa9a793aa/F2.jpg", "media/9c32ca83538fa9a793aa/F3.jpg", "media/9c32ca83538fa9a793aa/F4.jpg", "media/9c32ca83538fa9a793aa/F5.jpg", "media/9c32ca83538fa9a793aa/F6.jpg", "media/9c32ca83538fa9a793aa/F7.jpg", "media/9c32ca83538fa9a793aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f62f858ac71f25841e2d99a6092b5d6ea511ee53f17e08e85f77135990317ddf"}
{"qa_id": "MVS-Synth__0008__camera_view_direction_change__00", "video_id": "MVS-Synth__0008", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/fb0ac4d1978c771a822d/F0.jpg", "media/fb0ac4d1978c771a822d/F1.jpg", "media/fb0ac4d1978c771a822d/F2.jpg", "media/fb0ac4d1978c771a822d/F3.jpg", "media/fb0ac4d1978c771a822d/F4.jpg", "media/fb0ac4d1978c771a822d/F5.jpg", "media/fb0ac4d1978c771a822d/F6.jpg", "media/fb0ac4d1978c771a822d/F7.jpg", "media/fb0ac4d1978c771a822d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8c42e127caf576fe9581603a2bbd69f6900124ae7243982626c79ac570f2d51"}
{"qa_id": "MVS-Synth__0009__camera_translation_direction__00", "video_id": "MVS-Synth__0009", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/eda25c7adb642a416fdc/F0.jpg", "media/eda25c7adb642a416fdc/F1.jpg", "media/eda25c7adb642a416fdc/F2.jpg", "media/eda25c7adb642a416fdc/F3.jpg", "media/eda25c7adb642a416fdc/F4.jpg", "media/eda25c7adb642a416fdc/F5.jpg", "media/eda25c7adb642a416fdc/F6.jpg", "media/eda25c7adb642a416fdc/F7.jpg", "media/eda25c7adb642a416fdc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "169fd0cb27a56d6cc61a548e408263ea24f406e2530718848bb2ff51f616db09"}
{"qa_id": "MVS-Synth__0009__camera_view_direction_change__01", "video_id": "MVS-Synth__0009", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/eda25c7adb642a416fdc/F0.jpg", "media/eda25c7adb642a416fdc/F1.jpg", "media/eda25c7adb642a416fdc/F2.jpg", "media/eda25c7adb642a416fdc/F3.jpg", "media/eda25c7adb642a416fdc/F4.jpg", "media/eda25c7adb642a416fdc/F5.jpg", "media/eda25c7adb642a416fdc/F6.jpg", "media/eda25c7adb642a416fdc/F7.jpg", "media/eda25c7adb642a416fdc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "MVS-Synth__0010__camera_translation_direction__00", "video_id": "MVS-Synth__0010", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "forward", "backward"], "frame_paths": ["media/93db5e6b1cba41726d5f/F0.jpg", "media/93db5e6b1cba41726d5f/F1.jpg", "media/93db5e6b1cba41726d5f/F2.jpg", "media/93db5e6b1cba41726d5f/F3.jpg", "media/93db5e6b1cba41726d5f/F4.jpg", "media/93db5e6b1cba41726d5f/F5.jpg", "media/93db5e6b1cba41726d5f/F6.jpg", "media/93db5e6b1cba41726d5f/F7.jpg", "media/93db5e6b1cba41726d5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d115203d7643e723cce3ef83eedaa8f05a0e98e203707b6d28c6c490a28d28ac"}
{"qa_id": "MVS-Synth__0010__camera_view_direction_change__01", "video_id": "MVS-Synth__0010", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/93db5e6b1cba41726d5f/F0.jpg", "media/93db5e6b1cba41726d5f/F1.jpg", "media/93db5e6b1cba41726d5f/F2.jpg", "media/93db5e6b1cba41726d5f/F3.jpg", "media/93db5e6b1cba41726d5f/F4.jpg", "media/93db5e6b1cba41726d5f/F5.jpg", "media/93db5e6b1cba41726d5f/F6.jpg", "media/93db5e6b1cba41726d5f/F7.jpg", "media/93db5e6b1cba41726d5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5683ac2cba2889761f2107cb0cd0e4049566bdfcb4d4d2e9e175afe1d3d5de8"}
{"qa_id": "MVS-Synth__0011__camera_view_direction_change__00", "video_id": "MVS-Synth__0011", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/3e7c58703e4b727ac4dc/F0.jpg", "media/3e7c58703e4b727ac4dc/F1.jpg", "media/3e7c58703e4b727ac4dc/F2.jpg", "media/3e7c58703e4b727ac4dc/F3.jpg", "media/3e7c58703e4b727ac4dc/F4.jpg", "media/3e7c58703e4b727ac4dc/F5.jpg", "media/3e7c58703e4b727ac4dc/F6.jpg", "media/3e7c58703e4b727ac4dc/F7.jpg", "media/3e7c58703e4b727ac4dc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ef70746922c92c53aa339098679f211f93db7742fd522afe5b9c877f1ad5941b"}
{"qa_id": "MVS-Synth__0012__camera_view_direction_change__00", "video_id": "MVS-Synth__0012", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/2779d39ae3469549f70f/F0.jpg", "media/2779d39ae3469549f70f/F1.jpg", "media/2779d39ae3469549f70f/F2.jpg", "media/2779d39ae3469549f70f/F3.jpg", "media/2779d39ae3469549f70f/F4.jpg", "media/2779d39ae3469549f70f/F5.jpg", "media/2779d39ae3469549f70f/F6.jpg", "media/2779d39ae3469549f70f/F7.jpg", "media/2779d39ae3469549f70f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e5107715b4cc55de0adcfb3e914f29a1bf71fd0cb36f576f7d91d5c487723d0"}
{"qa_id": "MVS-Synth__0015__camera_translation_direction__00", "video_id": "MVS-Synth__0015", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/8b912b1662c2fbe9e92c/F0.jpg", "media/8b912b1662c2fbe9e92c/F1.jpg", "media/8b912b1662c2fbe9e92c/F2.jpg", "media/8b912b1662c2fbe9e92c/F3.jpg", "media/8b912b1662c2fbe9e92c/F4.jpg", "media/8b912b1662c2fbe9e92c/F5.jpg", "media/8b912b1662c2fbe9e92c/F6.jpg", "media/8b912b1662c2fbe9e92c/F7.jpg", "media/8b912b1662c2fbe9e92c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7d74971a991e0322fd67da7a50c6950e82ba322e4407956f96a7ab9928ef3cb"}
{"qa_id": "MVS-Synth__0015__camera_view_direction_change__01", "video_id": "MVS-Synth__0015", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/8b912b1662c2fbe9e92c/F0.jpg", "media/8b912b1662c2fbe9e92c/F1.jpg", "media/8b912b1662c2fbe9e92c/F2.jpg", "media/8b912b1662c2fbe9e92c/F3.jpg", "media/8b912b1662c2fbe9e92c/F4.jpg", "media/8b912b1662c2fbe9e92c/F5.jpg", "media/8b912b1662c2fbe9e92c/F6.jpg", "media/8b912b1662c2fbe9e92c/F7.jpg", "media/8b912b1662c2fbe9e92c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dfd5b7d4ffc4412cf6027ba6a85df51156eaa457b486f97f5d7bcc07cafdbd3e"}
{"qa_id": "MVS-Synth__0016__camera_translation_direction__00", "video_id": "MVS-Synth__0016", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/996922c6c78aa97d5c8c/F0.jpg", "media/996922c6c78aa97d5c8c/F1.jpg", "media/996922c6c78aa97d5c8c/F2.jpg", "media/996922c6c78aa97d5c8c/F3.jpg", "media/996922c6c78aa97d5c8c/F4.jpg", "media/996922c6c78aa97d5c8c/F5.jpg", "media/996922c6c78aa97d5c8c/F6.jpg", "media/996922c6c78aa97d5c8c/F7.jpg", "media/996922c6c78aa97d5c8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4b7f51dda1c5438550060e445cc5186c3a1f5c6715976603a3b3dac4cc2c3c38"}
{"qa_id": "MVS-Synth__0016__camera_view_direction_change__01", "video_id": "MVS-Synth__0016", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/996922c6c78aa97d5c8c/F0.jpg", "media/996922c6c78aa97d5c8c/F1.jpg", "media/996922c6c78aa97d5c8c/F2.jpg", "media/996922c6c78aa97d5c8c/F3.jpg", "media/996922c6c78aa97d5c8c/F4.jpg", "media/996922c6c78aa97d5c8c/F5.jpg", "media/996922c6c78aa97d5c8c/F6.jpg", "media/996922c6c78aa97d5c8c/F7.jpg", "media/996922c6c78aa97d5c8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f50dcc4587b168ab9459635029bd7a6e5fa68bafe0ef7289da83ccd89fc17f1f"}
{"qa_id": "MVS-Synth__0017__camera_translation_direction__00", "video_id": "MVS-Synth__0017", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/0dad5021c2c8caa0ed7e/F0.jpg", "media/0dad5021c2c8caa0ed7e/F1.jpg", "media/0dad5021c2c8caa0ed7e/F2.jpg", "media/0dad5021c2c8caa0ed7e/F3.jpg", "media/0dad5021c2c8caa0ed7e/F4.jpg", "media/0dad5021c2c8caa0ed7e/F5.jpg", "media/0dad5021c2c8caa0ed7e/F6.jpg", "media/0dad5021c2c8caa0ed7e/F7.jpg", "media/0dad5021c2c8caa0ed7e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "745353a39e0081ee36d03057eaa54f1abca2394ad575113152d306b0c122e5d9"}
{"qa_id": "MVS-Synth__0017__camera_view_direction_change__01", "video_id": "MVS-Synth__0017", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/0dad5021c2c8caa0ed7e/F0.jpg", "media/0dad5021c2c8caa0ed7e/F1.jpg", "media/0dad5021c2c8caa0ed7e/F2.jpg", "media/0dad5021c2c8caa0ed7e/F3.jpg", "media/0dad5021c2c8caa0ed7e/F4.jpg", "media/0dad5021c2c8caa0ed7e/F5.jpg", "media/0dad5021c2c8caa0ed7e/F6.jpg", "media/0dad5021c2c8caa0ed7e/F7.jpg", "media/0dad5021c2c8caa0ed7e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "81acf75395df3f91c9537db1dd7459bbce0f49cd99a9d1521c6f45ab3fea557c"}
{"qa_id": "MVS-Synth__0018__camera_view_direction_change__00", "video_id": "MVS-Synth__0018", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/9e868261f544ac59a8ff/F0.jpg", "media/9e868261f544ac59a8ff/F1.jpg", "media/9e868261f544ac59a8ff/F2.jpg", "media/9e868261f544ac59a8ff/F3.jpg", "media/9e868261f544ac59a8ff/F4.jpg", "media/9e868261f544ac59a8ff/F5.jpg", "media/9e868261f544ac59a8ff/F6.jpg", "media/9e868261f544ac59a8ff/F7.jpg", "media/9e868261f544ac59a8ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "919f1e79eafc227ca6dac1373423d12d3ec0017a6cf2734eba094df6cb34eb7b"}
{"qa_id": "MVS-Synth__0019__camera_translation_direction__00", "video_id": "MVS-Synth__0019", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/7dd82ba6d1019366391b/F0.jpg", "media/7dd82ba6d1019366391b/F1.jpg", "media/7dd82ba6d1019366391b/F2.jpg", "media/7dd82ba6d1019366391b/F3.jpg", "media/7dd82ba6d1019366391b/F4.jpg", "media/7dd82ba6d1019366391b/F5.jpg", "media/7dd82ba6d1019366391b/F6.jpg", "media/7dd82ba6d1019366391b/F7.jpg", "media/7dd82ba6d1019366391b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bcedde5222254971ac93c3b4c6b9e8520420583937da8f345a3a7bdaf8a90f8b"}
{"qa_id": "MVS-Synth__0019__camera_view_direction_change__01", "video_id": "MVS-Synth__0019", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/7dd82ba6d1019366391b/F0.jpg", "media/7dd82ba6d1019366391b/F1.jpg", "media/7dd82ba6d1019366391b/F2.jpg", "media/7dd82ba6d1019366391b/F3.jpg", "media/7dd82ba6d1019366391b/F4.jpg", "media/7dd82ba6d1019366391b/F5.jpg", "media/7dd82ba6d1019366391b/F6.jpg", "media/7dd82ba6d1019366391b/F7.jpg", "media/7dd82ba6d1019366391b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "MVS-Synth__0020__camera_view_direction_change__00", "video_id": "MVS-Synth__0020", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/abab2efd7e38ea35cb7a/F0.jpg", "media/abab2efd7e38ea35cb7a/F1.jpg", "media/abab2efd7e38ea35cb7a/F2.jpg", "media/abab2efd7e38ea35cb7a/F3.jpg", "media/abab2efd7e38ea35cb7a/F4.jpg", "media/abab2efd7e38ea35cb7a/F5.jpg", "media/abab2efd7e38ea35cb7a/F6.jpg", "media/abab2efd7e38ea35cb7a/F7.jpg", "media/abab2efd7e38ea35cb7a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b2aa5db618715fadd14639d9c93485d25797a205f11b90743eb79149be712a89"}
{"qa_id": "MVS-Synth__0021__camera_view_direction_change__00", "video_id": "MVS-Synth__0021", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/bedc879f748a07f66559/F0.jpg", "media/bedc879f748a07f66559/F1.jpg", "media/bedc879f748a07f66559/F2.jpg", "media/bedc879f748a07f66559/F3.jpg", "media/bedc879f748a07f66559/F4.jpg", "media/bedc879f748a07f66559/F5.jpg", "media/bedc879f748a07f66559/F6.jpg", "media/bedc879f748a07f66559/F7.jpg", "media/bedc879f748a07f66559/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74e7834987ad2bdda409db81a3760e1ca41987794e8ca3e25f0c1204c3290113"}
{"qa_id": "MVS-Synth__0022__camera_translation_direction__00", "video_id": "MVS-Synth__0022", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "right", "forward", "left"], "frame_paths": ["media/c4b0d67e0ad2a5404a0a/F0.jpg", "media/c4b0d67e0ad2a5404a0a/F1.jpg", "media/c4b0d67e0ad2a5404a0a/F2.jpg", "media/c4b0d67e0ad2a5404a0a/F3.jpg", "media/c4b0d67e0ad2a5404a0a/F4.jpg", "media/c4b0d67e0ad2a5404a0a/F5.jpg", "media/c4b0d67e0ad2a5404a0a/F6.jpg", "media/c4b0d67e0ad2a5404a0a/F7.jpg", "media/c4b0d67e0ad2a5404a0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "678add26b8f3c4523e0bba2c241b4980016cdd261b4cde1b9a167cb49df9b0df"}
{"qa_id": "MVS-Synth__0022__camera_view_direction_change__01", "video_id": "MVS-Synth__0022", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/c4b0d67e0ad2a5404a0a/F0.jpg", "media/c4b0d67e0ad2a5404a0a/F1.jpg", "media/c4b0d67e0ad2a5404a0a/F2.jpg", "media/c4b0d67e0ad2a5404a0a/F3.jpg", "media/c4b0d67e0ad2a5404a0a/F4.jpg", "media/c4b0d67e0ad2a5404a0a/F5.jpg", "media/c4b0d67e0ad2a5404a0a/F6.jpg", "media/c4b0d67e0ad2a5404a0a/F7.jpg", "media/c4b0d67e0ad2a5404a0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc4d7b0f5ea7a9d2e2e67f2004e27eb17ceb8e6eb8e91934ee2a004d65688d7f"}
{"qa_id": "MVS-Synth__0023__camera_translation_direction__00", "video_id": "MVS-Synth__0023", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "left", "down", "up"], "frame_paths": ["media/9b7440715c4879a02bcd/F0.jpg", "media/9b7440715c4879a02bcd/F1.jpg", "media/9b7440715c4879a02bcd/F2.jpg", "media/9b7440715c4879a02bcd/F3.jpg", "media/9b7440715c4879a02bcd/F4.jpg", "media/9b7440715c4879a02bcd/F5.jpg", "media/9b7440715c4879a02bcd/F6.jpg", "media/9b7440715c4879a02bcd/F7.jpg", "media/9b7440715c4879a02bcd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b375a8a37600205a4b1aea5436f8aac02c8feedda477478e60606831ceca7d7"}
{"qa_id": "MVS-Synth__0023__camera_view_direction_change__01", "video_id": "MVS-Synth__0023", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/9b7440715c4879a02bcd/F0.jpg", "media/9b7440715c4879a02bcd/F1.jpg", "media/9b7440715c4879a02bcd/F2.jpg", "media/9b7440715c4879a02bcd/F3.jpg", "media/9b7440715c4879a02bcd/F4.jpg", "media/9b7440715c4879a02bcd/F5.jpg", "media/9b7440715c4879a02bcd/F6.jpg", "media/9b7440715c4879a02bcd/F7.jpg", "media/9b7440715c4879a02bcd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aae452319b2f4945644936359ec7d3add997dd5675fbc7ea01e8e75c706f84dd"}
{"qa_id": "MVS-Synth__0024__camera_view_direction_change__00", "video_id": "MVS-Synth__0024", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/2c080d23adcfb529c682/F0.jpg", "media/2c080d23adcfb529c682/F1.jpg", "media/2c080d23adcfb529c682/F2.jpg", "media/2c080d23adcfb529c682/F3.jpg", "media/2c080d23adcfb529c682/F4.jpg", "media/2c080d23adcfb529c682/F5.jpg", "media/2c080d23adcfb529c682/F6.jpg", "media/2c080d23adcfb529c682/F7.jpg", "media/2c080d23adcfb529c682/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "786b10fdb9a3d0830fef254cf002e5fc91d843dc4ead29539b3fdef0fa0dffd4"}
{"qa_id": "MVS-Synth__0025__camera_translation_direction__00", "video_id": "MVS-Synth__0025", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/bc87290b55f652e54819/F0.jpg", "media/bc87290b55f652e54819/F1.jpg", "media/bc87290b55f652e54819/F2.jpg", "media/bc87290b55f652e54819/F3.jpg", "media/bc87290b55f652e54819/F4.jpg", "media/bc87290b55f652e54819/F5.jpg", "media/bc87290b55f652e54819/F6.jpg", "media/bc87290b55f652e54819/F7.jpg", "media/bc87290b55f652e54819/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1dc67968b7f2c8daea83f4e746d3c0b7be976a2b882797788660444396a9b32d"}
{"qa_id": "MVS-Synth__0025__camera_view_direction_change__01", "video_id": "MVS-Synth__0025", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/bc87290b55f652e54819/F0.jpg", "media/bc87290b55f652e54819/F1.jpg", "media/bc87290b55f652e54819/F2.jpg", "media/bc87290b55f652e54819/F3.jpg", "media/bc87290b55f652e54819/F4.jpg", "media/bc87290b55f652e54819/F5.jpg", "media/bc87290b55f652e54819/F6.jpg", "media/bc87290b55f652e54819/F7.jpg", "media/bc87290b55f652e54819/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e33a27c0df9fda77cf54a36a69c8e3817f698bb26735df3744d1f1d938f2c8a"}
{"qa_id": "MVS-Synth__0026__camera_translation_direction__00", "video_id": "MVS-Synth__0026", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/c6dac5dd75af3ac9cac1/F0.jpg", "media/c6dac5dd75af3ac9cac1/F1.jpg", "media/c6dac5dd75af3ac9cac1/F2.jpg", "media/c6dac5dd75af3ac9cac1/F3.jpg", "media/c6dac5dd75af3ac9cac1/F4.jpg", "media/c6dac5dd75af3ac9cac1/F5.jpg", "media/c6dac5dd75af3ac9cac1/F6.jpg", "media/c6dac5dd75af3ac9cac1/F7.jpg", "media/c6dac5dd75af3ac9cac1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7322cd8142b2a3ffa50cca56c8845e11fbd6b80edad6b6179c4a888dd51e0662"}
{"qa_id": "MVS-Synth__0026__camera_view_direction_change__01", "video_id": "MVS-Synth__0026", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/c6dac5dd75af3ac9cac1/F0.jpg", "media/c6dac5dd75af3ac9cac1/F1.jpg", "media/c6dac5dd75af3ac9cac1/F2.jpg", "media/c6dac5dd75af3ac9cac1/F3.jpg", "media/c6dac5dd75af3ac9cac1/F4.jpg", "media/c6dac5dd75af3ac9cac1/F5.jpg", "media/c6dac5dd75af3ac9cac1/F6.jpg", "media/c6dac5dd75af3ac9cac1/F7.jpg", "media/c6dac5dd75af3ac9cac1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f218a93d64b283f77cacb76b262f4772e24d2d79ec4ed3fc4cf74b94a6669fc9"}
{"qa_id": "MVS-Synth__0027__camera_translation_direction__00", "video_id": "MVS-Synth__0027", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/aa275217c0a10f543e8c/F0.jpg", "media/aa275217c0a10f543e8c/F1.jpg", "media/aa275217c0a10f543e8c/F2.jpg", "media/aa275217c0a10f543e8c/F3.jpg", "media/aa275217c0a10f543e8c/F4.jpg", "media/aa275217c0a10f543e8c/F5.jpg", "media/aa275217c0a10f543e8c/F6.jpg", "media/aa275217c0a10f543e8c/F7.jpg", "media/aa275217c0a10f543e8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a70fa241ae18f2b963b1ef6046e1f270fa824bbab42fc5617adbb1ae5678e137"}
{"qa_id": "MVS-Synth__0027__camera_view_direction_change__01", "video_id": "MVS-Synth__0027", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/aa275217c0a10f543e8c/F0.jpg", "media/aa275217c0a10f543e8c/F1.jpg", "media/aa275217c0a10f543e8c/F2.jpg", "media/aa275217c0a10f543e8c/F3.jpg", "media/aa275217c0a10f543e8c/F4.jpg", "media/aa275217c0a10f543e8c/F5.jpg", "media/aa275217c0a10f543e8c/F6.jpg", "media/aa275217c0a10f543e8c/F7.jpg", "media/aa275217c0a10f543e8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ccb648a24c1da4a8def80f074a4caac598b890f88c9c003e8f650c6387923239"}
{"qa_id": "MVS-Synth__0028__camera_view_direction_change__00", "video_id": "MVS-Synth__0028", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/6076ca7aa8758fb1dda9/F0.jpg", "media/6076ca7aa8758fb1dda9/F1.jpg", "media/6076ca7aa8758fb1dda9/F2.jpg", "media/6076ca7aa8758fb1dda9/F3.jpg", "media/6076ca7aa8758fb1dda9/F4.jpg", "media/6076ca7aa8758fb1dda9/F5.jpg", "media/6076ca7aa8758fb1dda9/F6.jpg", "media/6076ca7aa8758fb1dda9/F7.jpg", "media/6076ca7aa8758fb1dda9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f6694b6865c8c5fa2f30d9ba16c4058b9722ecf37893a078baf1bbf8519c41d"}
{"qa_id": "MVS-Synth__0029__camera_view_direction_change__00", "video_id": "MVS-Synth__0029", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/601aed0e3798fed935e6/F0.jpg", "media/601aed0e3798fed935e6/F1.jpg", "media/601aed0e3798fed935e6/F2.jpg", "media/601aed0e3798fed935e6/F3.jpg", "media/601aed0e3798fed935e6/F4.jpg", "media/601aed0e3798fed935e6/F5.jpg", "media/601aed0e3798fed935e6/F6.jpg", "media/601aed0e3798fed935e6/F7.jpg", "media/601aed0e3798fed935e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68990db0ed2196b37376443e0d7973cd6b92de97abb9c566854e3cdae4f1199b"}
{"qa_id": "MVS-Synth__0030__camera_translation_direction__00", "video_id": "MVS-Synth__0030", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/72596259aadafe8d2a18/F0.jpg", "media/72596259aadafe8d2a18/F1.jpg", "media/72596259aadafe8d2a18/F2.jpg", "media/72596259aadafe8d2a18/F3.jpg", "media/72596259aadafe8d2a18/F4.jpg", "media/72596259aadafe8d2a18/F5.jpg", "media/72596259aadafe8d2a18/F6.jpg", "media/72596259aadafe8d2a18/F7.jpg", "media/72596259aadafe8d2a18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31dee05e7529100cf636a9c811f18203f235fc63e4e139f19e2a462b90dfdf08"}
{"qa_id": "MVS-Synth__0031__camera_translation_direction__00", "video_id": "MVS-Synth__0031", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "right", "forward", "left"], "frame_paths": ["media/e1627f8d00f04b44c829/F0.jpg", "media/e1627f8d00f04b44c829/F1.jpg", "media/e1627f8d00f04b44c829/F2.jpg", "media/e1627f8d00f04b44c829/F3.jpg", "media/e1627f8d00f04b44c829/F4.jpg", "media/e1627f8d00f04b44c829/F5.jpg", "media/e1627f8d00f04b44c829/F6.jpg", "media/e1627f8d00f04b44c829/F7.jpg", "media/e1627f8d00f04b44c829/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "678add26b8f3c4523e0bba2c241b4980016cdd261b4cde1b9a167cb49df9b0df"}
{"qa_id": "MVS-Synth__0031__camera_view_direction_change__01", "video_id": "MVS-Synth__0031", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/e1627f8d00f04b44c829/F0.jpg", "media/e1627f8d00f04b44c829/F1.jpg", "media/e1627f8d00f04b44c829/F2.jpg", "media/e1627f8d00f04b44c829/F3.jpg", "media/e1627f8d00f04b44c829/F4.jpg", "media/e1627f8d00f04b44c829/F5.jpg", "media/e1627f8d00f04b44c829/F6.jpg", "media/e1627f8d00f04b44c829/F7.jpg", "media/e1627f8d00f04b44c829/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc4d7b0f5ea7a9d2e2e67f2004e27eb17ceb8e6eb8e91934ee2a004d65688d7f"}
{"qa_id": "MVS-Synth__0033__camera_view_direction_change__00", "video_id": "MVS-Synth__0033", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/ec0c3bc17c80bf24c2d1/F0.jpg", "media/ec0c3bc17c80bf24c2d1/F1.jpg", "media/ec0c3bc17c80bf24c2d1/F2.jpg", "media/ec0c3bc17c80bf24c2d1/F3.jpg", "media/ec0c3bc17c80bf24c2d1/F4.jpg", "media/ec0c3bc17c80bf24c2d1/F5.jpg", "media/ec0c3bc17c80bf24c2d1/F6.jpg", "media/ec0c3bc17c80bf24c2d1/F7.jpg", "media/ec0c3bc17c80bf24c2d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "973b87bd5357c095ba6aa15901cee4b9973a6a9dd3cc8d63fd02340a3998ef14"}
{"qa_id": "MVS-Synth__0034__camera_translation_direction__00", "video_id": "MVS-Synth__0034", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/2cecbef5350e6e58f15e/F0.jpg", "media/2cecbef5350e6e58f15e/F1.jpg", "media/2cecbef5350e6e58f15e/F2.jpg", "media/2cecbef5350e6e58f15e/F3.jpg", "media/2cecbef5350e6e58f15e/F4.jpg", "media/2cecbef5350e6e58f15e/F5.jpg", "media/2cecbef5350e6e58f15e/F6.jpg", "media/2cecbef5350e6e58f15e/F7.jpg", "media/2cecbef5350e6e58f15e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c27ba296f681b213cbfbdc54b52453af736800110e027012d25770fcd45ef701"}
{"qa_id": "MVS-Synth__0034__camera_view_direction_change__01", "video_id": "MVS-Synth__0034", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/2cecbef5350e6e58f15e/F0.jpg", "media/2cecbef5350e6e58f15e/F1.jpg", "media/2cecbef5350e6e58f15e/F2.jpg", "media/2cecbef5350e6e58f15e/F3.jpg", "media/2cecbef5350e6e58f15e/F4.jpg", "media/2cecbef5350e6e58f15e/F5.jpg", "media/2cecbef5350e6e58f15e/F6.jpg", "media/2cecbef5350e6e58f15e/F7.jpg", "media/2cecbef5350e6e58f15e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f6694b6865c8c5fa2f30d9ba16c4058b9722ecf37893a078baf1bbf8519c41d"}
{"qa_id": "MVS-Synth__0035__camera_translation_direction__00", "video_id": "MVS-Synth__0035", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/e0d02dfb1e0a8eb4ce86/F0.jpg", "media/e0d02dfb1e0a8eb4ce86/F1.jpg", "media/e0d02dfb1e0a8eb4ce86/F2.jpg", "media/e0d02dfb1e0a8eb4ce86/F3.jpg", "media/e0d02dfb1e0a8eb4ce86/F4.jpg", "media/e0d02dfb1e0a8eb4ce86/F5.jpg", "media/e0d02dfb1e0a8eb4ce86/F6.jpg", "media/e0d02dfb1e0a8eb4ce86/F7.jpg", "media/e0d02dfb1e0a8eb4ce86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c38dd11d63f137a8dde673263ddf3df83bc4bb83266aed0d14e2d465ca1d3cc"}
{"qa_id": "MVS-Synth__0035__camera_view_direction_change__01", "video_id": "MVS-Synth__0035", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/e0d02dfb1e0a8eb4ce86/F0.jpg", "media/e0d02dfb1e0a8eb4ce86/F1.jpg", "media/e0d02dfb1e0a8eb4ce86/F2.jpg", "media/e0d02dfb1e0a8eb4ce86/F3.jpg", "media/e0d02dfb1e0a8eb4ce86/F4.jpg", "media/e0d02dfb1e0a8eb4ce86/F5.jpg", "media/e0d02dfb1e0a8eb4ce86/F6.jpg", "media/e0d02dfb1e0a8eb4ce86/F7.jpg", "media/e0d02dfb1e0a8eb4ce86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "603e61a89b8b9a39b97e3ae385c877f99d6aded5272823baf22c4e4d39c96f7d"}
{"qa_id": "MVS-Synth__0036__camera_translation_direction__00", "video_id": "MVS-Synth__0036", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/21d55a56c3116530749d/F0.jpg", "media/21d55a56c3116530749d/F1.jpg", "media/21d55a56c3116530749d/F2.jpg", "media/21d55a56c3116530749d/F3.jpg", "media/21d55a56c3116530749d/F4.jpg", "media/21d55a56c3116530749d/F5.jpg", "media/21d55a56c3116530749d/F6.jpg", "media/21d55a56c3116530749d/F7.jpg", "media/21d55a56c3116530749d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a70fa241ae18f2b963b1ef6046e1f270fa824bbab42fc5617adbb1ae5678e137"}
{"qa_id": "MVS-Synth__0036__camera_view_direction_change__01", "video_id": "MVS-Synth__0036", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/21d55a56c3116530749d/F0.jpg", "media/21d55a56c3116530749d/F1.jpg", "media/21d55a56c3116530749d/F2.jpg", "media/21d55a56c3116530749d/F3.jpg", "media/21d55a56c3116530749d/F4.jpg", "media/21d55a56c3116530749d/F5.jpg", "media/21d55a56c3116530749d/F6.jpg", "media/21d55a56c3116530749d/F7.jpg", "media/21d55a56c3116530749d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b51f8251cd2ea30a949b0082ba091b815be44b540d789670cc7abe38e14beac9"}
{"qa_id": "MVS-Synth__0037__camera_translation_direction__00", "video_id": "MVS-Synth__0037", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/bc0d78d804628bc63cbb/F0.jpg", "media/bc0d78d804628bc63cbb/F1.jpg", "media/bc0d78d804628bc63cbb/F2.jpg", "media/bc0d78d804628bc63cbb/F3.jpg", "media/bc0d78d804628bc63cbb/F4.jpg", "media/bc0d78d804628bc63cbb/F5.jpg", "media/bc0d78d804628bc63cbb/F6.jpg", "media/bc0d78d804628bc63cbb/F7.jpg", "media/bc0d78d804628bc63cbb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d080008d41b6887855fb963da0ea1c5ee0d27812277b7b0fd39a8ac4004a1e7f"}
{"qa_id": "MVS-Synth__0037__camera_view_direction_change__01", "video_id": "MVS-Synth__0037", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/bc0d78d804628bc63cbb/F0.jpg", "media/bc0d78d804628bc63cbb/F1.jpg", "media/bc0d78d804628bc63cbb/F2.jpg", "media/bc0d78d804628bc63cbb/F3.jpg", "media/bc0d78d804628bc63cbb/F4.jpg", "media/bc0d78d804628bc63cbb/F5.jpg", "media/bc0d78d804628bc63cbb/F6.jpg", "media/bc0d78d804628bc63cbb/F7.jpg", "media/bc0d78d804628bc63cbb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "MVS-Synth__0038__camera_view_direction_change__00", "video_id": "MVS-Synth__0038", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/e12f5d5ed48a4ab2e8d5/F0.jpg", "media/e12f5d5ed48a4ab2e8d5/F1.jpg", "media/e12f5d5ed48a4ab2e8d5/F2.jpg", "media/e12f5d5ed48a4ab2e8d5/F3.jpg", "media/e12f5d5ed48a4ab2e8d5/F4.jpg", "media/e12f5d5ed48a4ab2e8d5/F5.jpg", "media/e12f5d5ed48a4ab2e8d5/F6.jpg", "media/e12f5d5ed48a4ab2e8d5/F7.jpg", "media/e12f5d5ed48a4ab2e8d5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d724f0d528a84374ef8ff2e618c158445dd772af363653bd9875d4b85c2bf99"}
{"qa_id": "MVS-Synth__0039__camera_translation_direction__00", "video_id": "MVS-Synth__0039", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/e748c17824de6edd3189/F0.jpg", "media/e748c17824de6edd3189/F1.jpg", "media/e748c17824de6edd3189/F2.jpg", "media/e748c17824de6edd3189/F3.jpg", "media/e748c17824de6edd3189/F4.jpg", "media/e748c17824de6edd3189/F5.jpg", "media/e748c17824de6edd3189/F6.jpg", "media/e748c17824de6edd3189/F7.jpg", "media/e748c17824de6edd3189/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "290f695252d7a79f9dac709e748bad27190745a119952270aa06e9bc38175e0d"}
{"qa_id": "MVS-Synth__0039__camera_view_direction_change__01", "video_id": "MVS-Synth__0039", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/e748c17824de6edd3189/F0.jpg", "media/e748c17824de6edd3189/F1.jpg", "media/e748c17824de6edd3189/F2.jpg", "media/e748c17824de6edd3189/F3.jpg", "media/e748c17824de6edd3189/F4.jpg", "media/e748c17824de6edd3189/F5.jpg", "media/e748c17824de6edd3189/F6.jpg", "media/e748c17824de6edd3189/F7.jpg", "media/e748c17824de6edd3189/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebaac16a469c69652194c2c56036315a21aa864aa41e667766e0e37841104248"}
{"qa_id": "MVS-Synth__0040__camera_translation_direction__00", "video_id": "MVS-Synth__0040", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "right", "forward", "left"], "frame_paths": ["media/9bcad1226fee942c22b7/F0.jpg", "media/9bcad1226fee942c22b7/F1.jpg", "media/9bcad1226fee942c22b7/F2.jpg", "media/9bcad1226fee942c22b7/F3.jpg", "media/9bcad1226fee942c22b7/F4.jpg", "media/9bcad1226fee942c22b7/F5.jpg", "media/9bcad1226fee942c22b7/F6.jpg", "media/9bcad1226fee942c22b7/F7.jpg", "media/9bcad1226fee942c22b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c96f7013dedb16568246c5e99cc1f4512ab741ac990ba070b9c77334e1090f9c"}
{"qa_id": "MVS-Synth__0040__camera_view_direction_change__01", "video_id": "MVS-Synth__0040", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/9bcad1226fee942c22b7/F0.jpg", "media/9bcad1226fee942c22b7/F1.jpg", "media/9bcad1226fee942c22b7/F2.jpg", "media/9bcad1226fee942c22b7/F3.jpg", "media/9bcad1226fee942c22b7/F4.jpg", "media/9bcad1226fee942c22b7/F5.jpg", "media/9bcad1226fee942c22b7/F6.jpg", "media/9bcad1226fee942c22b7/F7.jpg", "media/9bcad1226fee942c22b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1087f5041649a675286108fd2647c44d89864ac3d1ee2c0c2436ed349fb1c739"}
{"qa_id": "MVS-Synth__0042__camera_view_direction_change__00", "video_id": "MVS-Synth__0042", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/f99b1ccdcc06b5b9ec04/F0.jpg", "media/f99b1ccdcc06b5b9ec04/F1.jpg", "media/f99b1ccdcc06b5b9ec04/F2.jpg", "media/f99b1ccdcc06b5b9ec04/F3.jpg", "media/f99b1ccdcc06b5b9ec04/F4.jpg", "media/f99b1ccdcc06b5b9ec04/F5.jpg", "media/f99b1ccdcc06b5b9ec04/F6.jpg", "media/f99b1ccdcc06b5b9ec04/F7.jpg", "media/f99b1ccdcc06b5b9ec04/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa06e6069661e9dc2b88fb0b656b0d195fc4b50e527bbe95a905ac5eb368e5ce"}
{"qa_id": "MVS-Synth__0044__camera_translation_direction__00", "video_id": "MVS-Synth__0044", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/2b8a614ccaeabaaf6aae/F0.jpg", "media/2b8a614ccaeabaaf6aae/F1.jpg", "media/2b8a614ccaeabaaf6aae/F2.jpg", "media/2b8a614ccaeabaaf6aae/F3.jpg", "media/2b8a614ccaeabaaf6aae/F4.jpg", "media/2b8a614ccaeabaaf6aae/F5.jpg", "media/2b8a614ccaeabaaf6aae/F6.jpg", "media/2b8a614ccaeabaaf6aae/F7.jpg", "media/2b8a614ccaeabaaf6aae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b993a18763f6c945db011868ab095a0ffbb2ccd6041345039f38158d80857dbb"}
{"qa_id": "MVS-Synth__0044__camera_view_direction_change__01", "video_id": "MVS-Synth__0044", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/2b8a614ccaeabaaf6aae/F0.jpg", "media/2b8a614ccaeabaaf6aae/F1.jpg", "media/2b8a614ccaeabaaf6aae/F2.jpg", "media/2b8a614ccaeabaaf6aae/F3.jpg", "media/2b8a614ccaeabaaf6aae/F4.jpg", "media/2b8a614ccaeabaaf6aae/F5.jpg", "media/2b8a614ccaeabaaf6aae/F6.jpg", "media/2b8a614ccaeabaaf6aae/F7.jpg", "media/2b8a614ccaeabaaf6aae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b496d7e4e3374a4d50feaa2428f7524d83b8b404927d955b4ce613f278b8ad72"}
{"qa_id": "MVS-Synth__0046__camera_view_direction_change__00", "video_id": "MVS-Synth__0046", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/521729ae2ee75caae15c/F0.jpg", "media/521729ae2ee75caae15c/F1.jpg", "media/521729ae2ee75caae15c/F2.jpg", "media/521729ae2ee75caae15c/F3.jpg", "media/521729ae2ee75caae15c/F4.jpg", "media/521729ae2ee75caae15c/F5.jpg", "media/521729ae2ee75caae15c/F6.jpg", "media/521729ae2ee75caae15c/F7.jpg", "media/521729ae2ee75caae15c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a72d033d29a5ca3c54ff06f204dc5d1114b6c5d06000a8090a4f2874f67dde75"}
{"qa_id": "MVS-Synth__0047__camera_translation_direction__00", "video_id": "MVS-Synth__0047", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/4b3799c0053c1720aa03/F0.jpg", "media/4b3799c0053c1720aa03/F1.jpg", "media/4b3799c0053c1720aa03/F2.jpg", "media/4b3799c0053c1720aa03/F3.jpg", "media/4b3799c0053c1720aa03/F4.jpg", "media/4b3799c0053c1720aa03/F5.jpg", "media/4b3799c0053c1720aa03/F6.jpg", "media/4b3799c0053c1720aa03/F7.jpg", "media/4b3799c0053c1720aa03/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "539726ef021f37ef8f02bac0477fe1a9d93560afe3bbc0e580008be882053aea"}
{"qa_id": "MVS-Synth__0047__camera_view_direction_change__01", "video_id": "MVS-Synth__0047", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/4b3799c0053c1720aa03/F0.jpg", "media/4b3799c0053c1720aa03/F1.jpg", "media/4b3799c0053c1720aa03/F2.jpg", "media/4b3799c0053c1720aa03/F3.jpg", "media/4b3799c0053c1720aa03/F4.jpg", "media/4b3799c0053c1720aa03/F5.jpg", "media/4b3799c0053c1720aa03/F6.jpg", "media/4b3799c0053c1720aa03/F7.jpg", "media/4b3799c0053c1720aa03/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a16a06c6e2af53aa582d5aa53df816653b1c929e94504cc9184797e384013607"}
{"qa_id": "MVS-Synth__0049__camera_view_direction_change__00", "video_id": "MVS-Synth__0049", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/f9b4e1700700a54d272a/F0.jpg", "media/f9b4e1700700a54d272a/F1.jpg", "media/f9b4e1700700a54d272a/F2.jpg", "media/f9b4e1700700a54d272a/F3.jpg", "media/f9b4e1700700a54d272a/F4.jpg", "media/f9b4e1700700a54d272a/F5.jpg", "media/f9b4e1700700a54d272a/F6.jpg", "media/f9b4e1700700a54d272a/F7.jpg", "media/f9b4e1700700a54d272a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0037d65e5f18e9a8089f75779d3421f7f29bf009078d44d3a566737f3dfe6230"}
{"qa_id": "MVS-Synth__0050__camera_view_direction_change__00", "video_id": "MVS-Synth__0050", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/6006f60d1db988942574/F0.jpg", "media/6006f60d1db988942574/F1.jpg", "media/6006f60d1db988942574/F2.jpg", "media/6006f60d1db988942574/F3.jpg", "media/6006f60d1db988942574/F4.jpg", "media/6006f60d1db988942574/F5.jpg", "media/6006f60d1db988942574/F6.jpg", "media/6006f60d1db988942574/F7.jpg", "media/6006f60d1db988942574/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eddece1229e26f9e5036199adcff2d11746412fc240fefc79101d72f17f4e451"}
{"qa_id": "MVS-Synth__0051__camera_translation_direction__00", "video_id": "MVS-Synth__0051", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/01b3c1692f5dfc4162c7/F0.jpg", "media/01b3c1692f5dfc4162c7/F1.jpg", "media/01b3c1692f5dfc4162c7/F2.jpg", "media/01b3c1692f5dfc4162c7/F3.jpg", "media/01b3c1692f5dfc4162c7/F4.jpg", "media/01b3c1692f5dfc4162c7/F5.jpg", "media/01b3c1692f5dfc4162c7/F6.jpg", "media/01b3c1692f5dfc4162c7/F7.jpg", "media/01b3c1692f5dfc4162c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b412ac774cc58251645df09f623d9defab604b235ddcb72f7490b32f131bc95c"}
{"qa_id": "MVS-Synth__0051__camera_view_direction_change__01", "video_id": "MVS-Synth__0051", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/01b3c1692f5dfc4162c7/F0.jpg", "media/01b3c1692f5dfc4162c7/F1.jpg", "media/01b3c1692f5dfc4162c7/F2.jpg", "media/01b3c1692f5dfc4162c7/F3.jpg", "media/01b3c1692f5dfc4162c7/F4.jpg", "media/01b3c1692f5dfc4162c7/F5.jpg", "media/01b3c1692f5dfc4162c7/F6.jpg", "media/01b3c1692f5dfc4162c7/F7.jpg", "media/01b3c1692f5dfc4162c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7809c30ae0f1d21b4e882cd9ee222f30cccc0e1b8468e72a78dae4e48d0ee4e"}
{"qa_id": "MVS-Synth__0052__camera_translation_direction__00", "video_id": "MVS-Synth__0052", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/dde6e36c146b05e4470c/F0.jpg", "media/dde6e36c146b05e4470c/F1.jpg", "media/dde6e36c146b05e4470c/F2.jpg", "media/dde6e36c146b05e4470c/F3.jpg", "media/dde6e36c146b05e4470c/F4.jpg", "media/dde6e36c146b05e4470c/F5.jpg", "media/dde6e36c146b05e4470c/F6.jpg", "media/dde6e36c146b05e4470c/F7.jpg", "media/dde6e36c146b05e4470c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b8686680c77cd152da257885d5dae261a94354164bd1f26bbd72ae2e82334db9"}
{"qa_id": "MVS-Synth__0052__camera_view_direction_change__01", "video_id": "MVS-Synth__0052", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/dde6e36c146b05e4470c/F0.jpg", "media/dde6e36c146b05e4470c/F1.jpg", "media/dde6e36c146b05e4470c/F2.jpg", "media/dde6e36c146b05e4470c/F3.jpg", "media/dde6e36c146b05e4470c/F4.jpg", "media/dde6e36c146b05e4470c/F5.jpg", "media/dde6e36c146b05e4470c/F6.jpg", "media/dde6e36c146b05e4470c/F7.jpg", "media/dde6e36c146b05e4470c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c839b56480b0ace4e186348d86f787f046e97162f2fd9f012d81215b678136b1"}
{"qa_id": "MVS-Synth__0053__camera_translation_direction__00", "video_id": "MVS-Synth__0053", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/54ae3883ca2f28978270/F0.jpg", "media/54ae3883ca2f28978270/F1.jpg", "media/54ae3883ca2f28978270/F2.jpg", "media/54ae3883ca2f28978270/F3.jpg", "media/54ae3883ca2f28978270/F4.jpg", "media/54ae3883ca2f28978270/F5.jpg", "media/54ae3883ca2f28978270/F6.jpg", "media/54ae3883ca2f28978270/F7.jpg", "media/54ae3883ca2f28978270/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "baa6559791b0e1ca06892deec15f142fadef78466281f5df397d7dfcd34786cd"}
{"qa_id": "MVS-Synth__0053__camera_view_direction_change__01", "video_id": "MVS-Synth__0053", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/54ae3883ca2f28978270/F0.jpg", "media/54ae3883ca2f28978270/F1.jpg", "media/54ae3883ca2f28978270/F2.jpg", "media/54ae3883ca2f28978270/F3.jpg", "media/54ae3883ca2f28978270/F4.jpg", "media/54ae3883ca2f28978270/F5.jpg", "media/54ae3883ca2f28978270/F6.jpg", "media/54ae3883ca2f28978270/F7.jpg", "media/54ae3883ca2f28978270/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53eda269f6d9eb7cf97dc206d5e7093b2873b91fc6e510b9e39ce9e8cde770e"}
{"qa_id": "MVS-Synth__0054__camera_translation_direction__00", "video_id": "MVS-Synth__0054", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/da638e9bcc13cae4fc6f/F0.jpg", "media/da638e9bcc13cae4fc6f/F1.jpg", "media/da638e9bcc13cae4fc6f/F2.jpg", "media/da638e9bcc13cae4fc6f/F3.jpg", "media/da638e9bcc13cae4fc6f/F4.jpg", "media/da638e9bcc13cae4fc6f/F5.jpg", "media/da638e9bcc13cae4fc6f/F6.jpg", "media/da638e9bcc13cae4fc6f/F7.jpg", "media/da638e9bcc13cae4fc6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0f105a0481690422890b60e7eb4b50932aa13f5d9fddbdc9493afe7026fc1f79"}
{"qa_id": "MVS-Synth__0054__camera_view_direction_change__01", "video_id": "MVS-Synth__0054", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/da638e9bcc13cae4fc6f/F0.jpg", "media/da638e9bcc13cae4fc6f/F1.jpg", "media/da638e9bcc13cae4fc6f/F2.jpg", "media/da638e9bcc13cae4fc6f/F3.jpg", "media/da638e9bcc13cae4fc6f/F4.jpg", "media/da638e9bcc13cae4fc6f/F5.jpg", "media/da638e9bcc13cae4fc6f/F6.jpg", "media/da638e9bcc13cae4fc6f/F7.jpg", "media/da638e9bcc13cae4fc6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d600f647af09b05c01b84522a96e3e0a37dae5728a979dd03c4a9d2f97ac72e4"}
{"qa_id": "MVS-Synth__0055__camera_view_direction_change__00", "video_id": "MVS-Synth__0055", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/54884a8a0ee6509e7f7f/F0.jpg", "media/54884a8a0ee6509e7f7f/F1.jpg", "media/54884a8a0ee6509e7f7f/F2.jpg", "media/54884a8a0ee6509e7f7f/F3.jpg", "media/54884a8a0ee6509e7f7f/F4.jpg", "media/54884a8a0ee6509e7f7f/F5.jpg", "media/54884a8a0ee6509e7f7f/F6.jpg", "media/54884a8a0ee6509e7f7f/F7.jpg", "media/54884a8a0ee6509e7f7f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cff300f1f1c02d7cd62c3c6a7c35701b837678d874931fdf0a1ea32392a74068"}
{"qa_id": "MVS-Synth__0056__camera_translation_direction__00", "video_id": "MVS-Synth__0056", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/33a4318d67e12cc70d3d/F0.jpg", "media/33a4318d67e12cc70d3d/F1.jpg", "media/33a4318d67e12cc70d3d/F2.jpg", "media/33a4318d67e12cc70d3d/F3.jpg", "media/33a4318d67e12cc70d3d/F4.jpg", "media/33a4318d67e12cc70d3d/F5.jpg", "media/33a4318d67e12cc70d3d/F6.jpg", "media/33a4318d67e12cc70d3d/F7.jpg", "media/33a4318d67e12cc70d3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2c07c83df5926c8b07bdae3c4cd4af2ce530784c18d4c9c038b3b16d21d2003b"}
{"qa_id": "MVS-Synth__0056__camera_view_direction_change__01", "video_id": "MVS-Synth__0056", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/33a4318d67e12cc70d3d/F0.jpg", "media/33a4318d67e12cc70d3d/F1.jpg", "media/33a4318d67e12cc70d3d/F2.jpg", "media/33a4318d67e12cc70d3d/F3.jpg", "media/33a4318d67e12cc70d3d/F4.jpg", "media/33a4318d67e12cc70d3d/F5.jpg", "media/33a4318d67e12cc70d3d/F6.jpg", "media/33a4318d67e12cc70d3d/F7.jpg", "media/33a4318d67e12cc70d3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc84c23b9bc26f5c31d98452c54cf4b9be896a649775d150e6f5f9fd56bae203"}
{"qa_id": "MVS-Synth__0057__camera_view_direction_change__00", "video_id": "MVS-Synth__0057", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/b1eb1b993e2efdac94d2/F0.jpg", "media/b1eb1b993e2efdac94d2/F1.jpg", "media/b1eb1b993e2efdac94d2/F2.jpg", "media/b1eb1b993e2efdac94d2/F3.jpg", "media/b1eb1b993e2efdac94d2/F4.jpg", "media/b1eb1b993e2efdac94d2/F5.jpg", "media/b1eb1b993e2efdac94d2/F6.jpg", "media/b1eb1b993e2efdac94d2/F7.jpg", "media/b1eb1b993e2efdac94d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d0d73320bc523c85e596c60cbf8b23e836ea5ff75c57bdbe6cba8fbe6719d7a6"}
{"qa_id": "MVS-Synth__0058__camera_view_direction_change__00", "video_id": "MVS-Synth__0058", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/520eb49121667c5b608f/F0.jpg", "media/520eb49121667c5b608f/F1.jpg", "media/520eb49121667c5b608f/F2.jpg", "media/520eb49121667c5b608f/F3.jpg", "media/520eb49121667c5b608f/F4.jpg", "media/520eb49121667c5b608f/F5.jpg", "media/520eb49121667c5b608f/F6.jpg", "media/520eb49121667c5b608f/F7.jpg", "media/520eb49121667c5b608f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9e9b76767acadd54a6de2adfd760c5c5c5a40f6713b7b4e384d9c5a0b3abaa4"}
{"qa_id": "MVS-Synth__0059__camera_translation_direction__00", "video_id": "MVS-Synth__0059", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "down", "left", "forward"], "frame_paths": ["media/68ab6554d542376ebf63/F0.jpg", "media/68ab6554d542376ebf63/F1.jpg", "media/68ab6554d542376ebf63/F2.jpg", "media/68ab6554d542376ebf63/F3.jpg", "media/68ab6554d542376ebf63/F4.jpg", "media/68ab6554d542376ebf63/F5.jpg", "media/68ab6554d542376ebf63/F6.jpg", "media/68ab6554d542376ebf63/F7.jpg", "media/68ab6554d542376ebf63/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "574d881133531df7c41b84ca3ac335cd8aacaca17a58386ba8e64e4c2a365d82"}
{"qa_id": "MVS-Synth__0059__camera_view_direction_change__01", "video_id": "MVS-Synth__0059", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/68ab6554d542376ebf63/F0.jpg", "media/68ab6554d542376ebf63/F1.jpg", "media/68ab6554d542376ebf63/F2.jpg", "media/68ab6554d542376ebf63/F3.jpg", "media/68ab6554d542376ebf63/F4.jpg", "media/68ab6554d542376ebf63/F5.jpg", "media/68ab6554d542376ebf63/F6.jpg", "media/68ab6554d542376ebf63/F7.jpg", "media/68ab6554d542376ebf63/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dba736c77d24174e6757d4d08a62ed7cef1871fe8a9517600abd96322e91bf60"}
{"qa_id": "MVS-Synth__0060__camera_translation_direction__00", "video_id": "MVS-Synth__0060", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/e7477c9f7f454ca348d8/F0.jpg", "media/e7477c9f7f454ca348d8/F1.jpg", "media/e7477c9f7f454ca348d8/F2.jpg", "media/e7477c9f7f454ca348d8/F3.jpg", "media/e7477c9f7f454ca348d8/F4.jpg", "media/e7477c9f7f454ca348d8/F5.jpg", "media/e7477c9f7f454ca348d8/F6.jpg", "media/e7477c9f7f454ca348d8/F7.jpg", "media/e7477c9f7f454ca348d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f5d8734755420895aa60f7fc92ab9a8bb4a2929579f6b9eacc8c28e509dcb98"}
{"qa_id": "MVS-Synth__0060__camera_view_direction_change__01", "video_id": "MVS-Synth__0060", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/e7477c9f7f454ca348d8/F0.jpg", "media/e7477c9f7f454ca348d8/F1.jpg", "media/e7477c9f7f454ca348d8/F2.jpg", "media/e7477c9f7f454ca348d8/F3.jpg", "media/e7477c9f7f454ca348d8/F4.jpg", "media/e7477c9f7f454ca348d8/F5.jpg", "media/e7477c9f7f454ca348d8/F6.jpg", "media/e7477c9f7f454ca348d8/F7.jpg", "media/e7477c9f7f454ca348d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff82544052f24c713cffbe3db5af9941dc55c3c6643ccc2967e95534a1913dd0"}
{"qa_id": "MVS-Synth__0061__camera_translation_direction__00", "video_id": "MVS-Synth__0061", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/6c6a5dc15c78e9b78ac4/F0.jpg", "media/6c6a5dc15c78e9b78ac4/F1.jpg", "media/6c6a5dc15c78e9b78ac4/F2.jpg", "media/6c6a5dc15c78e9b78ac4/F3.jpg", "media/6c6a5dc15c78e9b78ac4/F4.jpg", "media/6c6a5dc15c78e9b78ac4/F5.jpg", "media/6c6a5dc15c78e9b78ac4/F6.jpg", "media/6c6a5dc15c78e9b78ac4/F7.jpg", "media/6c6a5dc15c78e9b78ac4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c27ba296f681b213cbfbdc54b52453af736800110e027012d25770fcd45ef701"}
{"qa_id": "MVS-Synth__0061__camera_view_direction_change__01", "video_id": "MVS-Synth__0061", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/6c6a5dc15c78e9b78ac4/F0.jpg", "media/6c6a5dc15c78e9b78ac4/F1.jpg", "media/6c6a5dc15c78e9b78ac4/F2.jpg", "media/6c6a5dc15c78e9b78ac4/F3.jpg", "media/6c6a5dc15c78e9b78ac4/F4.jpg", "media/6c6a5dc15c78e9b78ac4/F5.jpg", "media/6c6a5dc15c78e9b78ac4/F6.jpg", "media/6c6a5dc15c78e9b78ac4/F7.jpg", "media/6c6a5dc15c78e9b78ac4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a72d033d29a5ca3c54ff06f204dc5d1114b6c5d06000a8090a4f2874f67dde75"}
{"qa_id": "MVS-Synth__0062__camera_translation_direction__00", "video_id": "MVS-Synth__0062", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/e39494fea0af14195ca7/F0.jpg", "media/e39494fea0af14195ca7/F1.jpg", "media/e39494fea0af14195ca7/F2.jpg", "media/e39494fea0af14195ca7/F3.jpg", "media/e39494fea0af14195ca7/F4.jpg", "media/e39494fea0af14195ca7/F5.jpg", "media/e39494fea0af14195ca7/F6.jpg", "media/e39494fea0af14195ca7/F7.jpg", "media/e39494fea0af14195ca7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7130bb969ed0476c8c50ff12b61fee7dd1a2a2c820b5460bdee06238113ea088"}
{"qa_id": "MVS-Synth__0062__camera_view_direction_change__01", "video_id": "MVS-Synth__0062", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/e39494fea0af14195ca7/F0.jpg", "media/e39494fea0af14195ca7/F1.jpg", "media/e39494fea0af14195ca7/F2.jpg", "media/e39494fea0af14195ca7/F3.jpg", "media/e39494fea0af14195ca7/F4.jpg", "media/e39494fea0af14195ca7/F5.jpg", "media/e39494fea0af14195ca7/F6.jpg", "media/e39494fea0af14195ca7/F7.jpg", "media/e39494fea0af14195ca7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e075abf80b8f568c63df87a8c22d2a074ed86e6c7284c32de692757af209bfbd"}
{"qa_id": "MVS-Synth__0063__camera_view_direction_change__00", "video_id": "MVS-Synth__0063", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/c3ea158897425ddc9333/F0.jpg", "media/c3ea158897425ddc9333/F1.jpg", "media/c3ea158897425ddc9333/F2.jpg", "media/c3ea158897425ddc9333/F3.jpg", "media/c3ea158897425ddc9333/F4.jpg", "media/c3ea158897425ddc9333/F5.jpg", "media/c3ea158897425ddc9333/F6.jpg", "media/c3ea158897425ddc9333/F7.jpg", "media/c3ea158897425ddc9333/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9b6f465a3392bc24613165cd6cbefda0ad93d09b5e746c92b21aa1bcec7e7283"}
{"qa_id": "MVS-Synth__0064__camera_translation_direction__00", "video_id": "MVS-Synth__0064", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/512b7ee82ac4a22b55bc/F0.jpg", "media/512b7ee82ac4a22b55bc/F1.jpg", "media/512b7ee82ac4a22b55bc/F2.jpg", "media/512b7ee82ac4a22b55bc/F3.jpg", "media/512b7ee82ac4a22b55bc/F4.jpg", "media/512b7ee82ac4a22b55bc/F5.jpg", "media/512b7ee82ac4a22b55bc/F6.jpg", "media/512b7ee82ac4a22b55bc/F7.jpg", "media/512b7ee82ac4a22b55bc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7b115b6039da392acf5bb363b5d32c8489177c0cf3f7d6ee7a28678be1f0ce6c"}
{"qa_id": "MVS-Synth__0064__camera_view_direction_change__01", "video_id": "MVS-Synth__0064", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/512b7ee82ac4a22b55bc/F0.jpg", "media/512b7ee82ac4a22b55bc/F1.jpg", "media/512b7ee82ac4a22b55bc/F2.jpg", "media/512b7ee82ac4a22b55bc/F3.jpg", "media/512b7ee82ac4a22b55bc/F4.jpg", "media/512b7ee82ac4a22b55bc/F5.jpg", "media/512b7ee82ac4a22b55bc/F6.jpg", "media/512b7ee82ac4a22b55bc/F7.jpg", "media/512b7ee82ac4a22b55bc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "MVS-Synth__0065__camera_translation_direction__00", "video_id": "MVS-Synth__0065", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/d700b3631a4355d30bd6/F0.jpg", "media/d700b3631a4355d30bd6/F1.jpg", "media/d700b3631a4355d30bd6/F2.jpg", "media/d700b3631a4355d30bd6/F3.jpg", "media/d700b3631a4355d30bd6/F4.jpg", "media/d700b3631a4355d30bd6/F5.jpg", "media/d700b3631a4355d30bd6/F6.jpg", "media/d700b3631a4355d30bd6/F7.jpg", "media/d700b3631a4355d30bd6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "19c4338b520816bbf12e07bf4746b54e8053f4d409caac1fce15910a1fa12ffd"}
{"qa_id": "MVS-Synth__0065__camera_view_direction_change__01", "video_id": "MVS-Synth__0065", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/d700b3631a4355d30bd6/F0.jpg", "media/d700b3631a4355d30bd6/F1.jpg", "media/d700b3631a4355d30bd6/F2.jpg", "media/d700b3631a4355d30bd6/F3.jpg", "media/d700b3631a4355d30bd6/F4.jpg", "media/d700b3631a4355d30bd6/F5.jpg", "media/d700b3631a4355d30bd6/F6.jpg", "media/d700b3631a4355d30bd6/F7.jpg", "media/d700b3631a4355d30bd6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5d47aef5ded49513cb794135dfce74960c5515464cb16527eec00058b2c9cfe"}
{"qa_id": "MVS-Synth__0066__camera_view_direction_change__00", "video_id": "MVS-Synth__0066", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/0a722e1d5045f2c9e983/F0.jpg", "media/0a722e1d5045f2c9e983/F1.jpg", "media/0a722e1d5045f2c9e983/F2.jpg", "media/0a722e1d5045f2c9e983/F3.jpg", "media/0a722e1d5045f2c9e983/F4.jpg", "media/0a722e1d5045f2c9e983/F5.jpg", "media/0a722e1d5045f2c9e983/F6.jpg", "media/0a722e1d5045f2c9e983/F7.jpg", "media/0a722e1d5045f2c9e983/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "06c33645116852296f667c9f60d17c910442097c29c5492688a07dbb53526843"}
{"qa_id": "MVS-Synth__0068__camera_view_direction_change__00", "video_id": "MVS-Synth__0068", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/d83915da1499fbddd935/F0.jpg", "media/d83915da1499fbddd935/F1.jpg", "media/d83915da1499fbddd935/F2.jpg", "media/d83915da1499fbddd935/F3.jpg", "media/d83915da1499fbddd935/F4.jpg", "media/d83915da1499fbddd935/F5.jpg", "media/d83915da1499fbddd935/F6.jpg", "media/d83915da1499fbddd935/F7.jpg", "media/d83915da1499fbddd935/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "973b87bd5357c095ba6aa15901cee4b9973a6a9dd3cc8d63fd02340a3998ef14"}
{"qa_id": "MVS-Synth__0069__camera_translation_direction__00", "video_id": "MVS-Synth__0069", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/6682e16d05d90d373dd2/F0.jpg", "media/6682e16d05d90d373dd2/F1.jpg", "media/6682e16d05d90d373dd2/F2.jpg", "media/6682e16d05d90d373dd2/F3.jpg", "media/6682e16d05d90d373dd2/F4.jpg", "media/6682e16d05d90d373dd2/F5.jpg", "media/6682e16d05d90d373dd2/F6.jpg", "media/6682e16d05d90d373dd2/F7.jpg", "media/6682e16d05d90d373dd2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "008399c6ca344e6cb95d204277c3f9662d8804dfa0c1dce362453b1460c4325b"}
{"qa_id": "MVS-Synth__0069__camera_view_direction_change__01", "video_id": "MVS-Synth__0069", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/6682e16d05d90d373dd2/F0.jpg", "media/6682e16d05d90d373dd2/F1.jpg", "media/6682e16d05d90d373dd2/F2.jpg", "media/6682e16d05d90d373dd2/F3.jpg", "media/6682e16d05d90d373dd2/F4.jpg", "media/6682e16d05d90d373dd2/F5.jpg", "media/6682e16d05d90d373dd2/F6.jpg", "media/6682e16d05d90d373dd2/F7.jpg", "media/6682e16d05d90d373dd2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73de12a0740d8ac42d7f504c3eb8957260a8ec07133febede7f4e1a1219726ad"}
{"qa_id": "MVS-Synth__0070__camera_translation_direction__00", "video_id": "MVS-Synth__0070", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "backward", "left", "right"], "frame_paths": ["media/7943be4e2756a0130c45/F0.jpg", "media/7943be4e2756a0130c45/F1.jpg", "media/7943be4e2756a0130c45/F2.jpg", "media/7943be4e2756a0130c45/F3.jpg", "media/7943be4e2756a0130c45/F4.jpg", "media/7943be4e2756a0130c45/F5.jpg", "media/7943be4e2756a0130c45/F6.jpg", "media/7943be4e2756a0130c45/F7.jpg", "media/7943be4e2756a0130c45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "644fa870e3b9b6a7652c58c2340fa23ffa5cb5631c5b1d8d304f6ba52dc8713c"}
{"qa_id": "MVS-Synth__0070__camera_view_direction_change__01", "video_id": "MVS-Synth__0070", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/7943be4e2756a0130c45/F0.jpg", "media/7943be4e2756a0130c45/F1.jpg", "media/7943be4e2756a0130c45/F2.jpg", "media/7943be4e2756a0130c45/F3.jpg", "media/7943be4e2756a0130c45/F4.jpg", "media/7943be4e2756a0130c45/F5.jpg", "media/7943be4e2756a0130c45/F6.jpg", "media/7943be4e2756a0130c45/F7.jpg", "media/7943be4e2756a0130c45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f98011f6f5e3bef7856bffbee3a8c2579537b239284b48df76fd5bbadf4bc3f"}
{"qa_id": "MVS-Synth__0072__camera_view_direction_change__00", "video_id": "MVS-Synth__0072", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/fa39a9f58dafd1e58e32/F0.jpg", "media/fa39a9f58dafd1e58e32/F1.jpg", "media/fa39a9f58dafd1e58e32/F2.jpg", "media/fa39a9f58dafd1e58e32/F3.jpg", "media/fa39a9f58dafd1e58e32/F4.jpg", "media/fa39a9f58dafd1e58e32/F5.jpg", "media/fa39a9f58dafd1e58e32/F6.jpg", "media/fa39a9f58dafd1e58e32/F7.jpg", "media/fa39a9f58dafd1e58e32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53eda269f6d9eb7cf97dc206d5e7093b2873b91fc6e510b9e39ce9e8cde770e"}
{"qa_id": "MVS-Synth__0073__camera_translation_direction__00", "video_id": "MVS-Synth__0073", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/54f2cb240250b9f6a3c5/F0.jpg", "media/54f2cb240250b9f6a3c5/F1.jpg", "media/54f2cb240250b9f6a3c5/F2.jpg", "media/54f2cb240250b9f6a3c5/F3.jpg", "media/54f2cb240250b9f6a3c5/F4.jpg", "media/54f2cb240250b9f6a3c5/F5.jpg", "media/54f2cb240250b9f6a3c5/F6.jpg", "media/54f2cb240250b9f6a3c5/F7.jpg", "media/54f2cb240250b9f6a3c5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7cabbaa9bbbf13197f27116d11be7df6399251121e811cdc14887b762741c6d"}
{"qa_id": "MVS-Synth__0073__camera_view_direction_change__01", "video_id": "MVS-Synth__0073", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/54f2cb240250b9f6a3c5/F0.jpg", "media/54f2cb240250b9f6a3c5/F1.jpg", "media/54f2cb240250b9f6a3c5/F2.jpg", "media/54f2cb240250b9f6a3c5/F3.jpg", "media/54f2cb240250b9f6a3c5/F4.jpg", "media/54f2cb240250b9f6a3c5/F5.jpg", "media/54f2cb240250b9f6a3c5/F6.jpg", "media/54f2cb240250b9f6a3c5/F7.jpg", "media/54f2cb240250b9f6a3c5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0992c9a7c7ac161bba93e87487b0832cb9226f187491df7b7358b9b65732bb4d"}
{"qa_id": "MVS-Synth__0074__camera_translation_direction__00", "video_id": "MVS-Synth__0074", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/780364e4165824161b44/F0.jpg", "media/780364e4165824161b44/F1.jpg", "media/780364e4165824161b44/F2.jpg", "media/780364e4165824161b44/F3.jpg", "media/780364e4165824161b44/F4.jpg", "media/780364e4165824161b44/F5.jpg", "media/780364e4165824161b44/F6.jpg", "media/780364e4165824161b44/F7.jpg", "media/780364e4165824161b44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2c07c83df5926c8b07bdae3c4cd4af2ce530784c18d4c9c038b3b16d21d2003b"}
{"qa_id": "MVS-Synth__0074__camera_view_direction_change__01", "video_id": "MVS-Synth__0074", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/780364e4165824161b44/F0.jpg", "media/780364e4165824161b44/F1.jpg", "media/780364e4165824161b44/F2.jpg", "media/780364e4165824161b44/F3.jpg", "media/780364e4165824161b44/F4.jpg", "media/780364e4165824161b44/F5.jpg", "media/780364e4165824161b44/F6.jpg", "media/780364e4165824161b44/F7.jpg", "media/780364e4165824161b44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eb3226fa28b7cbaa59a774008af1d74c7020689b3ec43515b1c459600bedb13a"}
{"qa_id": "MVS-Synth__0075__camera_translation_direction__00", "video_id": "MVS-Synth__0075", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/a26fd8240a8d4a4c5d95/F0.jpg", "media/a26fd8240a8d4a4c5d95/F1.jpg", "media/a26fd8240a8d4a4c5d95/F2.jpg", "media/a26fd8240a8d4a4c5d95/F3.jpg", "media/a26fd8240a8d4a4c5d95/F4.jpg", "media/a26fd8240a8d4a4c5d95/F5.jpg", "media/a26fd8240a8d4a4c5d95/F6.jpg", "media/a26fd8240a8d4a4c5d95/F7.jpg", "media/a26fd8240a8d4a4c5d95/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "290f695252d7a79f9dac709e748bad27190745a119952270aa06e9bc38175e0d"}
{"qa_id": "MVS-Synth__0075__camera_view_direction_change__01", "video_id": "MVS-Synth__0075", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/a26fd8240a8d4a4c5d95/F0.jpg", "media/a26fd8240a8d4a4c5d95/F1.jpg", "media/a26fd8240a8d4a4c5d95/F2.jpg", "media/a26fd8240a8d4a4c5d95/F3.jpg", "media/a26fd8240a8d4a4c5d95/F4.jpg", "media/a26fd8240a8d4a4c5d95/F5.jpg", "media/a26fd8240a8d4a4c5d95/F6.jpg", "media/a26fd8240a8d4a4c5d95/F7.jpg", "media/a26fd8240a8d4a4c5d95/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c62e4cce265ee0250831b207eb7336b04a7d9ae9f67b14a17d5a22e813943899"}
{"qa_id": "MVS-Synth__0076__camera_translation_direction__00", "video_id": "MVS-Synth__0076", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "forward", "left", "down"], "frame_paths": ["media/4a2cd5d60053b858cf97/F0.jpg", "media/4a2cd5d60053b858cf97/F1.jpg", "media/4a2cd5d60053b858cf97/F2.jpg", "media/4a2cd5d60053b858cf97/F3.jpg", "media/4a2cd5d60053b858cf97/F4.jpg", "media/4a2cd5d60053b858cf97/F5.jpg", "media/4a2cd5d60053b858cf97/F6.jpg", "media/4a2cd5d60053b858cf97/F7.jpg", "media/4a2cd5d60053b858cf97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0236d104382b9f38766083ab14a14f9a66b2493366cf216ee55dc04dd0b242c9"}
{"qa_id": "MVS-Synth__0076__camera_view_direction_change__01", "video_id": "MVS-Synth__0076", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/4a2cd5d60053b858cf97/F0.jpg", "media/4a2cd5d60053b858cf97/F1.jpg", "media/4a2cd5d60053b858cf97/F2.jpg", "media/4a2cd5d60053b858cf97/F3.jpg", "media/4a2cd5d60053b858cf97/F4.jpg", "media/4a2cd5d60053b858cf97/F5.jpg", "media/4a2cd5d60053b858cf97/F6.jpg", "media/4a2cd5d60053b858cf97/F7.jpg", "media/4a2cd5d60053b858cf97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "973b87bd5357c095ba6aa15901cee4b9973a6a9dd3cc8d63fd02340a3998ef14"}
{"qa_id": "MVS-Synth__0077__camera_translation_direction__00", "video_id": "MVS-Synth__0077", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "down", "left", "forward"], "frame_paths": ["media/afb564fe1ba666d5dea0/F0.jpg", "media/afb564fe1ba666d5dea0/F1.jpg", "media/afb564fe1ba666d5dea0/F2.jpg", "media/afb564fe1ba666d5dea0/F3.jpg", "media/afb564fe1ba666d5dea0/F4.jpg", "media/afb564fe1ba666d5dea0/F5.jpg", "media/afb564fe1ba666d5dea0/F6.jpg", "media/afb564fe1ba666d5dea0/F7.jpg", "media/afb564fe1ba666d5dea0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dde19642f74e4f0e8c614fd48bb2a0793478becbbc632a49514f490471bccf42"}
{"qa_id": "MVS-Synth__0078__camera_view_direction_change__00", "video_id": "MVS-Synth__0078", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/93b00977fd7da76bc67d/F0.jpg", "media/93b00977fd7da76bc67d/F1.jpg", "media/93b00977fd7da76bc67d/F2.jpg", "media/93b00977fd7da76bc67d/F3.jpg", "media/93b00977fd7da76bc67d/F4.jpg", "media/93b00977fd7da76bc67d/F5.jpg", "media/93b00977fd7da76bc67d/F6.jpg", "media/93b00977fd7da76bc67d/F7.jpg", "media/93b00977fd7da76bc67d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbb87362fd2146075129df0a63ca048b7c2b3853d76ff3874fec52e8e219a8f7"}
{"qa_id": "MVS-Synth__0079__camera_translation_direction__00", "video_id": "MVS-Synth__0079", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "up", "forward", "left"], "frame_paths": ["media/72a3a9bf73487274d1ee/F0.jpg", "media/72a3a9bf73487274d1ee/F1.jpg", "media/72a3a9bf73487274d1ee/F2.jpg", "media/72a3a9bf73487274d1ee/F3.jpg", "media/72a3a9bf73487274d1ee/F4.jpg", "media/72a3a9bf73487274d1ee/F5.jpg", "media/72a3a9bf73487274d1ee/F6.jpg", "media/72a3a9bf73487274d1ee/F7.jpg", "media/72a3a9bf73487274d1ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b891c926b9b4401c267ad0dc5aeb25ff48298e6f91a8b0a832ce848f269ac350"}
{"qa_id": "MVS-Synth__0079__camera_view_direction_change__01", "video_id": "MVS-Synth__0079", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/72a3a9bf73487274d1ee/F0.jpg", "media/72a3a9bf73487274d1ee/F1.jpg", "media/72a3a9bf73487274d1ee/F2.jpg", "media/72a3a9bf73487274d1ee/F3.jpg", "media/72a3a9bf73487274d1ee/F4.jpg", "media/72a3a9bf73487274d1ee/F5.jpg", "media/72a3a9bf73487274d1ee/F6.jpg", "media/72a3a9bf73487274d1ee/F7.jpg", "media/72a3a9bf73487274d1ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eb3226fa28b7cbaa59a774008af1d74c7020689b3ec43515b1c459600bedb13a"}
{"qa_id": "MVS-Synth__0080__camera_translation_direction__00", "video_id": "MVS-Synth__0080", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/d943c1a3b035b61c24e7/F0.jpg", "media/d943c1a3b035b61c24e7/F1.jpg", "media/d943c1a3b035b61c24e7/F2.jpg", "media/d943c1a3b035b61c24e7/F3.jpg", "media/d943c1a3b035b61c24e7/F4.jpg", "media/d943c1a3b035b61c24e7/F5.jpg", "media/d943c1a3b035b61c24e7/F6.jpg", "media/d943c1a3b035b61c24e7/F7.jpg", "media/d943c1a3b035b61c24e7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7130bb969ed0476c8c50ff12b61fee7dd1a2a2c820b5460bdee06238113ea088"}
{"qa_id": "MVS-Synth__0080__camera_view_direction_change__01", "video_id": "MVS-Synth__0080", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/d943c1a3b035b61c24e7/F0.jpg", "media/d943c1a3b035b61c24e7/F1.jpg", "media/d943c1a3b035b61c24e7/F2.jpg", "media/d943c1a3b035b61c24e7/F3.jpg", "media/d943c1a3b035b61c24e7/F4.jpg", "media/d943c1a3b035b61c24e7/F5.jpg", "media/d943c1a3b035b61c24e7/F6.jpg", "media/d943c1a3b035b61c24e7/F7.jpg", "media/d943c1a3b035b61c24e7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e075abf80b8f568c63df87a8c22d2a074ed86e6c7284c32de692757af209bfbd"}
{"qa_id": "MVS-Synth__0081__camera_translation_direction__00", "video_id": "MVS-Synth__0081", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/2388f092b4af50fae52d/F0.jpg", "media/2388f092b4af50fae52d/F1.jpg", "media/2388f092b4af50fae52d/F2.jpg", "media/2388f092b4af50fae52d/F3.jpg", "media/2388f092b4af50fae52d/F4.jpg", "media/2388f092b4af50fae52d/F5.jpg", "media/2388f092b4af50fae52d/F6.jpg", "media/2388f092b4af50fae52d/F7.jpg", "media/2388f092b4af50fae52d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4abdfe4e0668e94f7c937504e0ef2a9ff3dcc62742617a52d7f9acaad2e42ae5"}
{"qa_id": "MVS-Synth__0081__camera_view_direction_change__01", "video_id": "MVS-Synth__0081", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/2388f092b4af50fae52d/F0.jpg", "media/2388f092b4af50fae52d/F1.jpg", "media/2388f092b4af50fae52d/F2.jpg", "media/2388f092b4af50fae52d/F3.jpg", "media/2388f092b4af50fae52d/F4.jpg", "media/2388f092b4af50fae52d/F5.jpg", "media/2388f092b4af50fae52d/F6.jpg", "media/2388f092b4af50fae52d/F7.jpg", "media/2388f092b4af50fae52d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e33a27c0df9fda77cf54a36a69c8e3817f698bb26735df3744d1f1d938f2c8a"}
{"qa_id": "MVS-Synth__0083__camera_translation_direction__00", "video_id": "MVS-Synth__0083", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/b31ca9ea0f860270f868/F0.jpg", "media/b31ca9ea0f860270f868/F1.jpg", "media/b31ca9ea0f860270f868/F2.jpg", "media/b31ca9ea0f860270f868/F3.jpg", "media/b31ca9ea0f860270f868/F4.jpg", "media/b31ca9ea0f860270f868/F5.jpg", "media/b31ca9ea0f860270f868/F6.jpg", "media/b31ca9ea0f860270f868/F7.jpg", "media/b31ca9ea0f860270f868/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "792357820a3748074038ff6dd3dd269a070da5cf6c31b2c37cf93cb69c3a497c"}
{"qa_id": "MVS-Synth__0083__camera_view_direction_change__01", "video_id": "MVS-Synth__0083", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/b31ca9ea0f860270f868/F0.jpg", "media/b31ca9ea0f860270f868/F1.jpg", "media/b31ca9ea0f860270f868/F2.jpg", "media/b31ca9ea0f860270f868/F3.jpg", "media/b31ca9ea0f860270f868/F4.jpg", "media/b31ca9ea0f860270f868/F5.jpg", "media/b31ca9ea0f860270f868/F6.jpg", "media/b31ca9ea0f860270f868/F7.jpg", "media/b31ca9ea0f860270f868/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5497e2cae4f1fb0f1552022fb7513a2cd0e85254f951200dc391ea03e3ae309"}
{"qa_id": "MVS-Synth__0084__camera_translation_direction__00", "video_id": "MVS-Synth__0084", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/0dc1de04cb8488d95c3d/F0.jpg", "media/0dc1de04cb8488d95c3d/F1.jpg", "media/0dc1de04cb8488d95c3d/F2.jpg", "media/0dc1de04cb8488d95c3d/F3.jpg", "media/0dc1de04cb8488d95c3d/F4.jpg", "media/0dc1de04cb8488d95c3d/F5.jpg", "media/0dc1de04cb8488d95c3d/F6.jpg", "media/0dc1de04cb8488d95c3d/F7.jpg", "media/0dc1de04cb8488d95c3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "09f8365dfe91ae59148d0d0e8267fedd13dafe3870f4dedafed5be44f318a651"}
{"qa_id": "MVS-Synth__0084__camera_view_direction_change__01", "video_id": "MVS-Synth__0084", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/0dc1de04cb8488d95c3d/F0.jpg", "media/0dc1de04cb8488d95c3d/F1.jpg", "media/0dc1de04cb8488d95c3d/F2.jpg", "media/0dc1de04cb8488d95c3d/F3.jpg", "media/0dc1de04cb8488d95c3d/F4.jpg", "media/0dc1de04cb8488d95c3d/F5.jpg", "media/0dc1de04cb8488d95c3d/F6.jpg", "media/0dc1de04cb8488d95c3d/F7.jpg", "media/0dc1de04cb8488d95c3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9e9b76767acadd54a6de2adfd760c5c5c5a40f6713b7b4e384d9c5a0b3abaa4"}
{"qa_id": "MVS-Synth__0085__camera_translation_direction__00", "video_id": "MVS-Synth__0085", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "backward", "forward", "left", "down"], "frame_paths": ["media/ee39e43a85818fa020b7/F0.jpg", "media/ee39e43a85818fa020b7/F1.jpg", "media/ee39e43a85818fa020b7/F2.jpg", "media/ee39e43a85818fa020b7/F3.jpg", "media/ee39e43a85818fa020b7/F4.jpg", "media/ee39e43a85818fa020b7/F5.jpg", "media/ee39e43a85818fa020b7/F6.jpg", "media/ee39e43a85818fa020b7/F7.jpg", "media/ee39e43a85818fa020b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. backward\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb46546d0b7b0aebe517f4d21540d1ab5aa13fad2a5e889779f6795c55a3beb7"}
{"qa_id": "MVS-Synth__0085__camera_view_direction_change__01", "video_id": "MVS-Synth__0085", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/ee39e43a85818fa020b7/F0.jpg", "media/ee39e43a85818fa020b7/F1.jpg", "media/ee39e43a85818fa020b7/F2.jpg", "media/ee39e43a85818fa020b7/F3.jpg", "media/ee39e43a85818fa020b7/F4.jpg", "media/ee39e43a85818fa020b7/F5.jpg", "media/ee39e43a85818fa020b7/F6.jpg", "media/ee39e43a85818fa020b7/F7.jpg", "media/ee39e43a85818fa020b7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a8dcf419b2189dc47199e6d66c32d6c180cca2c6fafb6a2a3c4339783c0dcf93"}
{"qa_id": "MVS-Synth__0086__camera_translation_direction__00", "video_id": "MVS-Synth__0086", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "down", "left", "forward"], "frame_paths": ["media/eb740d831f111c086e9c/F0.jpg", "media/eb740d831f111c086e9c/F1.jpg", "media/eb740d831f111c086e9c/F2.jpg", "media/eb740d831f111c086e9c/F3.jpg", "media/eb740d831f111c086e9c/F4.jpg", "media/eb740d831f111c086e9c/F5.jpg", "media/eb740d831f111c086e9c/F6.jpg", "media/eb740d831f111c086e9c/F7.jpg", "media/eb740d831f111c086e9c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "92ef72b4c89978d3285139954cf390690a4c30f869c385ae08afe38cbd28d157"}
{"qa_id": "MVS-Synth__0086__camera_view_direction_change__01", "video_id": "MVS-Synth__0086", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/eb740d831f111c086e9c/F0.jpg", "media/eb740d831f111c086e9c/F1.jpg", "media/eb740d831f111c086e9c/F2.jpg", "media/eb740d831f111c086e9c/F3.jpg", "media/eb740d831f111c086e9c/F4.jpg", "media/eb740d831f111c086e9c/F5.jpg", "media/eb740d831f111c086e9c/F6.jpg", "media/eb740d831f111c086e9c/F7.jpg", "media/eb740d831f111c086e9c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d018176bea3ca3cbb86de6353c7fe991172ad8416d3f9fa330e36569a35405d"}
{"qa_id": "MVS-Synth__0087__camera_translation_direction__00", "video_id": "MVS-Synth__0087", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/34e9cd50163bf94bdcb5/F0.jpg", "media/34e9cd50163bf94bdcb5/F1.jpg", "media/34e9cd50163bf94bdcb5/F2.jpg", "media/34e9cd50163bf94bdcb5/F3.jpg", "media/34e9cd50163bf94bdcb5/F4.jpg", "media/34e9cd50163bf94bdcb5/F5.jpg", "media/34e9cd50163bf94bdcb5/F6.jpg", "media/34e9cd50163bf94bdcb5/F7.jpg", "media/34e9cd50163bf94bdcb5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36911c11d99ecc728b350513d7be52a045ff7eb03300d6b518d1525da76c08b8"}
{"qa_id": "MVS-Synth__0087__camera_view_direction_change__01", "video_id": "MVS-Synth__0087", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/34e9cd50163bf94bdcb5/F0.jpg", "media/34e9cd50163bf94bdcb5/F1.jpg", "media/34e9cd50163bf94bdcb5/F2.jpg", "media/34e9cd50163bf94bdcb5/F3.jpg", "media/34e9cd50163bf94bdcb5/F4.jpg", "media/34e9cd50163bf94bdcb5/F5.jpg", "media/34e9cd50163bf94bdcb5/F6.jpg", "media/34e9cd50163bf94bdcb5/F7.jpg", "media/34e9cd50163bf94bdcb5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a1e80f61da85c51436c41da1489606dcec6ae5cf6271135a94de2cf2d2c5a02"}
{"qa_id": "MVS-Synth__0088__camera_view_direction_change__00", "video_id": "MVS-Synth__0088", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/c79c9750058b8fefd1a1/F0.jpg", "media/c79c9750058b8fefd1a1/F1.jpg", "media/c79c9750058b8fefd1a1/F2.jpg", "media/c79c9750058b8fefd1a1/F3.jpg", "media/c79c9750058b8fefd1a1/F4.jpg", "media/c79c9750058b8fefd1a1/F5.jpg", "media/c79c9750058b8fefd1a1/F6.jpg", "media/c79c9750058b8fefd1a1/F7.jpg", "media/c79c9750058b8fefd1a1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ec7d27631c2fb63ac4b3f9ba4485ae932dd511329f8841700414a6418a04b6c4"}
{"qa_id": "MVS-Synth__0089__camera_translation_direction__00", "video_id": "MVS-Synth__0089", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "forward", "up", "left"], "frame_paths": ["media/7ccaeeb1afdcc3e45ebb/F0.jpg", "media/7ccaeeb1afdcc3e45ebb/F1.jpg", "media/7ccaeeb1afdcc3e45ebb/F2.jpg", "media/7ccaeeb1afdcc3e45ebb/F3.jpg", "media/7ccaeeb1afdcc3e45ebb/F4.jpg", "media/7ccaeeb1afdcc3e45ebb/F5.jpg", "media/7ccaeeb1afdcc3e45ebb/F6.jpg", "media/7ccaeeb1afdcc3e45ebb/F7.jpg", "media/7ccaeeb1afdcc3e45ebb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb11f7ed33ebdab6f9bb2b33c0c25902cb85cde8874ff87f4a06fdfc0fc83d6"}
{"qa_id": "MVS-Synth__0090__camera_translation_direction__00", "video_id": "MVS-Synth__0090", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/97c72cf178e9e71be21b/F0.jpg", "media/97c72cf178e9e71be21b/F1.jpg", "media/97c72cf178e9e71be21b/F2.jpg", "media/97c72cf178e9e71be21b/F3.jpg", "media/97c72cf178e9e71be21b/F4.jpg", "media/97c72cf178e9e71be21b/F5.jpg", "media/97c72cf178e9e71be21b/F6.jpg", "media/97c72cf178e9e71be21b/F7.jpg", "media/97c72cf178e9e71be21b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "169fd0cb27a56d6cc61a548e408263ea24f406e2530718848bb2ff51f616db09"}
{"qa_id": "MVS-Synth__0090__camera_view_direction_change__01", "video_id": "MVS-Synth__0090", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/97c72cf178e9e71be21b/F0.jpg", "media/97c72cf178e9e71be21b/F1.jpg", "media/97c72cf178e9e71be21b/F2.jpg", "media/97c72cf178e9e71be21b/F3.jpg", "media/97c72cf178e9e71be21b/F4.jpg", "media/97c72cf178e9e71be21b/F5.jpg", "media/97c72cf178e9e71be21b/F6.jpg", "media/97c72cf178e9e71be21b/F7.jpg", "media/97c72cf178e9e71be21b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8c42e127caf576fe9581603a2bbd69f6900124ae7243982626c79ac570f2d51"}
{"qa_id": "MVS-Synth__0091__camera_view_direction_change__00", "video_id": "MVS-Synth__0091", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/8b9d74386f212a7e93a8/F0.jpg", "media/8b9d74386f212a7e93a8/F1.jpg", "media/8b9d74386f212a7e93a8/F2.jpg", "media/8b9d74386f212a7e93a8/F3.jpg", "media/8b9d74386f212a7e93a8/F4.jpg", "media/8b9d74386f212a7e93a8/F5.jpg", "media/8b9d74386f212a7e93a8/F6.jpg", "media/8b9d74386f212a7e93a8/F7.jpg", "media/8b9d74386f212a7e93a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f6694b6865c8c5fa2f30d9ba16c4058b9722ecf37893a078baf1bbf8519c41d"}
{"qa_id": "MVS-Synth__0092__camera_view_direction_change__00", "video_id": "MVS-Synth__0092", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/2a7d6b9512f104341a5c/F0.jpg", "media/2a7d6b9512f104341a5c/F1.jpg", "media/2a7d6b9512f104341a5c/F2.jpg", "media/2a7d6b9512f104341a5c/F3.jpg", "media/2a7d6b9512f104341a5c/F4.jpg", "media/2a7d6b9512f104341a5c/F5.jpg", "media/2a7d6b9512f104341a5c/F6.jpg", "media/2a7d6b9512f104341a5c/F7.jpg", "media/2a7d6b9512f104341a5c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "182c5aa79ac692be900eb9ea77fb24bbb6f34e7c3218c28f9ae2a0ad2f38658c"}
{"qa_id": "MVS-Synth__0093__camera_view_direction_change__00", "video_id": "MVS-Synth__0093", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/07e096e737927b6fa9b3/F0.jpg", "media/07e096e737927b6fa9b3/F1.jpg", "media/07e096e737927b6fa9b3/F2.jpg", "media/07e096e737927b6fa9b3/F3.jpg", "media/07e096e737927b6fa9b3/F4.jpg", "media/07e096e737927b6fa9b3/F5.jpg", "media/07e096e737927b6fa9b3/F6.jpg", "media/07e096e737927b6fa9b3/F7.jpg", "media/07e096e737927b6fa9b3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5f9d57aba1726663a6995bfde2e966d089c85ecba4b9a7459139f285c8114db"}
{"qa_id": "MVS-Synth__0094__camera_view_direction_change__00", "video_id": "MVS-Synth__0094", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/20ec2457f60cc3b936ab/F0.jpg", "media/20ec2457f60cc3b936ab/F1.jpg", "media/20ec2457f60cc3b936ab/F2.jpg", "media/20ec2457f60cc3b936ab/F3.jpg", "media/20ec2457f60cc3b936ab/F4.jpg", "media/20ec2457f60cc3b936ab/F5.jpg", "media/20ec2457f60cc3b936ab/F6.jpg", "media/20ec2457f60cc3b936ab/F7.jpg", "media/20ec2457f60cc3b936ab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ffb9fe8324fc9c1e31db5d6b73f7684a41bc94d6dafddadb04f4e802205742ba"}
{"qa_id": "MVS-Synth__0095__camera_view_direction_change__00", "video_id": "MVS-Synth__0095", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/bfa362051e7e22b137f0/F0.jpg", "media/bfa362051e7e22b137f0/F1.jpg", "media/bfa362051e7e22b137f0/F2.jpg", "media/bfa362051e7e22b137f0/F3.jpg", "media/bfa362051e7e22b137f0/F4.jpg", "media/bfa362051e7e22b137f0/F5.jpg", "media/bfa362051e7e22b137f0/F6.jpg", "media/bfa362051e7e22b137f0/F7.jpg", "media/bfa362051e7e22b137f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd5b55e0cf9c57a555ae502873a107b4eead5311b4ee8834dce0d37fe23cc30b"}
{"qa_id": "MVS-Synth__0096__camera_view_direction_change__00", "video_id": "MVS-Synth__0096", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/e637fbc4464be6c98092/F0.jpg", "media/e637fbc4464be6c98092/F1.jpg", "media/e637fbc4464be6c98092/F2.jpg", "media/e637fbc4464be6c98092/F3.jpg", "media/e637fbc4464be6c98092/F4.jpg", "media/e637fbc4464be6c98092/F5.jpg", "media/e637fbc4464be6c98092/F6.jpg", "media/e637fbc4464be6c98092/F7.jpg", "media/e637fbc4464be6c98092/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1e44244a3cf4d0c52f7ff5873c26b746b6943e4684ccbff5d65a33f2f2bb1922"}
{"qa_id": "MVS-Synth__0097__camera_view_direction_change__00", "video_id": "MVS-Synth__0097", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/5190678f7e5d7785097b/F0.jpg", "media/5190678f7e5d7785097b/F1.jpg", "media/5190678f7e5d7785097b/F2.jpg", "media/5190678f7e5d7785097b/F3.jpg", "media/5190678f7e5d7785097b/F4.jpg", "media/5190678f7e5d7785097b/F5.jpg", "media/5190678f7e5d7785097b/F6.jpg", "media/5190678f7e5d7785097b/F7.jpg", "media/5190678f7e5d7785097b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "93fd8df73d31bc8e34ac6bdb52cfdf36ae7d1d71294b0a1e86c1abe14ef61687"}
{"qa_id": "MVS-Synth__0098__camera_view_direction_change__00", "video_id": "MVS-Synth__0098", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/13e9603ad51fb3a96a89/F0.jpg", "media/13e9603ad51fb3a96a89/F1.jpg", "media/13e9603ad51fb3a96a89/F2.jpg", "media/13e9603ad51fb3a96a89/F3.jpg", "media/13e9603ad51fb3a96a89/F4.jpg", "media/13e9603ad51fb3a96a89/F5.jpg", "media/13e9603ad51fb3a96a89/F6.jpg", "media/13e9603ad51fb3a96a89/F7.jpg", "media/13e9603ad51fb3a96a89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a16a06c6e2af53aa582d5aa53df816653b1c929e94504cc9184797e384013607"}
{"qa_id": "MVS-Synth__0100__camera_translation_direction__00", "video_id": "MVS-Synth__0100", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "right", "down", "up", "forward", "backward"], "frame_paths": ["media/5f93ea7ffdc6cc57e10e/F0.jpg", "media/5f93ea7ffdc6cc57e10e/F1.jpg", "media/5f93ea7ffdc6cc57e10e/F2.jpg", "media/5f93ea7ffdc6cc57e10e/F3.jpg", "media/5f93ea7ffdc6cc57e10e/F4.jpg", "media/5f93ea7ffdc6cc57e10e/F5.jpg", "media/5f93ea7ffdc6cc57e10e/F6.jpg", "media/5f93ea7ffdc6cc57e10e/F7.jpg", "media/5f93ea7ffdc6cc57e10e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. down\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "04a3540e89d362e734de1ace9f7a58dd7dcf72a01cc7d89c45740cb4d69ac23c"}
{"qa_id": "MVS-Synth__0100__camera_view_direction_change__01", "video_id": "MVS-Synth__0100", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/5f93ea7ffdc6cc57e10e/F0.jpg", "media/5f93ea7ffdc6cc57e10e/F1.jpg", "media/5f93ea7ffdc6cc57e10e/F2.jpg", "media/5f93ea7ffdc6cc57e10e/F3.jpg", "media/5f93ea7ffdc6cc57e10e/F4.jpg", "media/5f93ea7ffdc6cc57e10e/F5.jpg", "media/5f93ea7ffdc6cc57e10e/F6.jpg", "media/5f93ea7ffdc6cc57e10e/F7.jpg", "media/5f93ea7ffdc6cc57e10e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a7b7d4beaabaf33ce877c2a21dcbf5ac3b60b372923a88c5d385cb625bfadc3"}
{"qa_id": "MVS-Synth__0101__camera_view_direction_change__00", "video_id": "MVS-Synth__0101", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/590bdab790defcc93a06/F0.jpg", "media/590bdab790defcc93a06/F1.jpg", "media/590bdab790defcc93a06/F2.jpg", "media/590bdab790defcc93a06/F3.jpg", "media/590bdab790defcc93a06/F4.jpg", "media/590bdab790defcc93a06/F5.jpg", "media/590bdab790defcc93a06/F6.jpg", "media/590bdab790defcc93a06/F7.jpg", "media/590bdab790defcc93a06/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "338bf0d2a239eb0e07fce4e6acff633b5fb0550211a45f41d3473c1c3d83a74e"}
{"qa_id": "MVS-Synth__0102__camera_view_direction_change__00", "video_id": "MVS-Synth__0102", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/f1237a2dcfe718eb6e40/F0.jpg", "media/f1237a2dcfe718eb6e40/F1.jpg", "media/f1237a2dcfe718eb6e40/F2.jpg", "media/f1237a2dcfe718eb6e40/F3.jpg", "media/f1237a2dcfe718eb6e40/F4.jpg", "media/f1237a2dcfe718eb6e40/F5.jpg", "media/f1237a2dcfe718eb6e40/F6.jpg", "media/f1237a2dcfe718eb6e40/F7.jpg", "media/f1237a2dcfe718eb6e40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "872ff7eaf0b3dabf6b2cc2ec86d5c3160888ee4506d0162427e06f0b1044c757"}
{"qa_id": "MVS-Synth__0103__camera_translation_direction__00", "video_id": "MVS-Synth__0103", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "right", "forward", "left"], "frame_paths": ["media/e8f24c168bc33ff9158b/F0.jpg", "media/e8f24c168bc33ff9158b/F1.jpg", "media/e8f24c168bc33ff9158b/F2.jpg", "media/e8f24c168bc33ff9158b/F3.jpg", "media/e8f24c168bc33ff9158b/F4.jpg", "media/e8f24c168bc33ff9158b/F5.jpg", "media/e8f24c168bc33ff9158b/F6.jpg", "media/e8f24c168bc33ff9158b/F7.jpg", "media/e8f24c168bc33ff9158b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "69890aace1b13b3e646b680a38b59bb0d43c243efa6366b91f7e5f077cc51adf"}
{"qa_id": "MVS-Synth__0103__camera_view_direction_change__01", "video_id": "MVS-Synth__0103", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/e8f24c168bc33ff9158b/F0.jpg", "media/e8f24c168bc33ff9158b/F1.jpg", "media/e8f24c168bc33ff9158b/F2.jpg", "media/e8f24c168bc33ff9158b/F3.jpg", "media/e8f24c168bc33ff9158b/F4.jpg", "media/e8f24c168bc33ff9158b/F5.jpg", "media/e8f24c168bc33ff9158b/F6.jpg", "media/e8f24c168bc33ff9158b/F7.jpg", "media/e8f24c168bc33ff9158b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eddece1229e26f9e5036199adcff2d11746412fc240fefc79101d72f17f4e451"}
{"qa_id": "MVS-Synth__0104__camera_view_direction_change__00", "video_id": "MVS-Synth__0104", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/cb4c3f7ae591a19971b2/F0.jpg", "media/cb4c3f7ae591a19971b2/F1.jpg", "media/cb4c3f7ae591a19971b2/F2.jpg", "media/cb4c3f7ae591a19971b2/F3.jpg", "media/cb4c3f7ae591a19971b2/F4.jpg", "media/cb4c3f7ae591a19971b2/F5.jpg", "media/cb4c3f7ae591a19971b2/F6.jpg", "media/cb4c3f7ae591a19971b2/F7.jpg", "media/cb4c3f7ae591a19971b2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb4db86f6ad842455c849665523a4999ea4c48a30c8629c691fd80577c13536b"}
{"qa_id": "MVS-Synth__0105__camera_translation_direction__00", "video_id": "MVS-Synth__0105", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/ab2ac7015760e96b8502/F0.jpg", "media/ab2ac7015760e96b8502/F1.jpg", "media/ab2ac7015760e96b8502/F2.jpg", "media/ab2ac7015760e96b8502/F3.jpg", "media/ab2ac7015760e96b8502/F4.jpg", "media/ab2ac7015760e96b8502/F5.jpg", "media/ab2ac7015760e96b8502/F6.jpg", "media/ab2ac7015760e96b8502/F7.jpg", "media/ab2ac7015760e96b8502/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f5d8734755420895aa60f7fc92ab9a8bb4a2929579f6b9eacc8c28e509dcb98"}
{"qa_id": "MVS-Synth__0105__camera_view_direction_change__01", "video_id": "MVS-Synth__0105", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/ab2ac7015760e96b8502/F0.jpg", "media/ab2ac7015760e96b8502/F1.jpg", "media/ab2ac7015760e96b8502/F2.jpg", "media/ab2ac7015760e96b8502/F3.jpg", "media/ab2ac7015760e96b8502/F4.jpg", "media/ab2ac7015760e96b8502/F5.jpg", "media/ab2ac7015760e96b8502/F6.jpg", "media/ab2ac7015760e96b8502/F7.jpg", "media/ab2ac7015760e96b8502/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff82544052f24c713cffbe3db5af9941dc55c3c6643ccc2967e95534a1913dd0"}
{"qa_id": "MVS-Synth__0106__camera_view_direction_change__00", "video_id": "MVS-Synth__0106", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/c99aee525a1715c90013/F0.jpg", "media/c99aee525a1715c90013/F1.jpg", "media/c99aee525a1715c90013/F2.jpg", "media/c99aee525a1715c90013/F3.jpg", "media/c99aee525a1715c90013/F4.jpg", "media/c99aee525a1715c90013/F5.jpg", "media/c99aee525a1715c90013/F6.jpg", "media/c99aee525a1715c90013/F7.jpg", "media/c99aee525a1715c90013/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "31811c288c54e3781c4c209e7681d6ddac10f01e1ba79dc08a3cbfbe6960a5b9"}
{"qa_id": "MVS-Synth__0107__camera_translation_direction__00", "video_id": "MVS-Synth__0107", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/1d0871e2dbf54b18e49d/F0.jpg", "media/1d0871e2dbf54b18e49d/F1.jpg", "media/1d0871e2dbf54b18e49d/F2.jpg", "media/1d0871e2dbf54b18e49d/F3.jpg", "media/1d0871e2dbf54b18e49d/F4.jpg", "media/1d0871e2dbf54b18e49d/F5.jpg", "media/1d0871e2dbf54b18e49d/F6.jpg", "media/1d0871e2dbf54b18e49d/F7.jpg", "media/1d0871e2dbf54b18e49d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "baa6559791b0e1ca06892deec15f142fadef78466281f5df397d7dfcd34786cd"}
{"qa_id": "MVS-Synth__0107__camera_view_direction_change__01", "video_id": "MVS-Synth__0107", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/1d0871e2dbf54b18e49d/F0.jpg", "media/1d0871e2dbf54b18e49d/F1.jpg", "media/1d0871e2dbf54b18e49d/F2.jpg", "media/1d0871e2dbf54b18e49d/F3.jpg", "media/1d0871e2dbf54b18e49d/F4.jpg", "media/1d0871e2dbf54b18e49d/F5.jpg", "media/1d0871e2dbf54b18e49d/F6.jpg", "media/1d0871e2dbf54b18e49d/F7.jpg", "media/1d0871e2dbf54b18e49d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53eda269f6d9eb7cf97dc206d5e7093b2873b91fc6e510b9e39ce9e8cde770e"}
{"qa_id": "MVS-Synth__0108__camera_view_direction_change__00", "video_id": "MVS-Synth__0108", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/96110f0a7019e9347de9/F0.jpg", "media/96110f0a7019e9347de9/F1.jpg", "media/96110f0a7019e9347de9/F2.jpg", "media/96110f0a7019e9347de9/F3.jpg", "media/96110f0a7019e9347de9/F4.jpg", "media/96110f0a7019e9347de9/F5.jpg", "media/96110f0a7019e9347de9/F6.jpg", "media/96110f0a7019e9347de9/F7.jpg", "media/96110f0a7019e9347de9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a2b981cf58cc18aed7896cfde47fa4ff64dca71be5067e9bea95e9bb65a23862"}
{"qa_id": "MVS-Synth__0109__camera_translation_direction__00", "video_id": "MVS-Synth__0109", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/55d4a429fe99d04011dd/F0.jpg", "media/55d4a429fe99d04011dd/F1.jpg", "media/55d4a429fe99d04011dd/F2.jpg", "media/55d4a429fe99d04011dd/F3.jpg", "media/55d4a429fe99d04011dd/F4.jpg", "media/55d4a429fe99d04011dd/F5.jpg", "media/55d4a429fe99d04011dd/F6.jpg", "media/55d4a429fe99d04011dd/F7.jpg", "media/55d4a429fe99d04011dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f439afcf9a0eff9637042cd5f6fdc372a1f2d98e6659a7ccb8188f30f3826330"}
{"qa_id": "MVS-Synth__0109__camera_view_direction_change__01", "video_id": "MVS-Synth__0109", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/55d4a429fe99d04011dd/F0.jpg", "media/55d4a429fe99d04011dd/F1.jpg", "media/55d4a429fe99d04011dd/F2.jpg", "media/55d4a429fe99d04011dd/F3.jpg", "media/55d4a429fe99d04011dd/F4.jpg", "media/55d4a429fe99d04011dd/F5.jpg", "media/55d4a429fe99d04011dd/F6.jpg", "media/55d4a429fe99d04011dd/F7.jpg", "media/55d4a429fe99d04011dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "MVS-Synth__0110__camera_translation_direction__00", "video_id": "MVS-Synth__0110", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "left", "up", "forward"], "frame_paths": ["media/c6f51f7135e09435fd86/F0.jpg", "media/c6f51f7135e09435fd86/F1.jpg", "media/c6f51f7135e09435fd86/F2.jpg", "media/c6f51f7135e09435fd86/F3.jpg", "media/c6f51f7135e09435fd86/F4.jpg", "media/c6f51f7135e09435fd86/F5.jpg", "media/c6f51f7135e09435fd86/F6.jpg", "media/c6f51f7135e09435fd86/F7.jpg", "media/c6f51f7135e09435fd86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc06b22651f650d389be185b2694d9832dc2622195791726fe83c2eb3b1bf315"}
{"qa_id": "MVS-Synth__0110__camera_view_direction_change__01", "video_id": "MVS-Synth__0110", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/c6f51f7135e09435fd86/F0.jpg", "media/c6f51f7135e09435fd86/F1.jpg", "media/c6f51f7135e09435fd86/F2.jpg", "media/c6f51f7135e09435fd86/F3.jpg", "media/c6f51f7135e09435fd86/F4.jpg", "media/c6f51f7135e09435fd86/F5.jpg", "media/c6f51f7135e09435fd86/F6.jpg", "media/c6f51f7135e09435fd86/F7.jpg", "media/c6f51f7135e09435fd86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74e7834987ad2bdda409db81a3760e1ca41987794e8ca3e25f0c1204c3290113"}
{"qa_id": "MVS-Synth__0112__camera_view_direction_change__00", "video_id": "MVS-Synth__0112", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/5ca994e0bc695c86a3f1/F0.jpg", "media/5ca994e0bc695c86a3f1/F1.jpg", "media/5ca994e0bc695c86a3f1/F2.jpg", "media/5ca994e0bc695c86a3f1/F3.jpg", "media/5ca994e0bc695c86a3f1/F4.jpg", "media/5ca994e0bc695c86a3f1/F5.jpg", "media/5ca994e0bc695c86a3f1/F6.jpg", "media/5ca994e0bc695c86a3f1/F7.jpg", "media/5ca994e0bc695c86a3f1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13cde86d65bc22544cfc80714a977a427a5635139a29b705b9b737262acbd489"}
{"qa_id": "MVS-Synth__0113__camera_view_direction_change__00", "video_id": "MVS-Synth__0113", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/049ecda8f68eff9e3ed4/F0.jpg", "media/049ecda8f68eff9e3ed4/F1.jpg", "media/049ecda8f68eff9e3ed4/F2.jpg", "media/049ecda8f68eff9e3ed4/F3.jpg", "media/049ecda8f68eff9e3ed4/F4.jpg", "media/049ecda8f68eff9e3ed4/F5.jpg", "media/049ecda8f68eff9e3ed4/F6.jpg", "media/049ecda8f68eff9e3ed4/F7.jpg", "media/049ecda8f68eff9e3ed4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "737208658eae7a6a71270b4761bf2e11d47bd90157a9455db3ef055f59607304"}
{"qa_id": "MVS-Synth__0114__camera_view_direction_change__00", "video_id": "MVS-Synth__0114", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/0ae1e1413171c6ddb65a/F0.jpg", "media/0ae1e1413171c6ddb65a/F1.jpg", "media/0ae1e1413171c6ddb65a/F2.jpg", "media/0ae1e1413171c6ddb65a/F3.jpg", "media/0ae1e1413171c6ddb65a/F4.jpg", "media/0ae1e1413171c6ddb65a/F5.jpg", "media/0ae1e1413171c6ddb65a/F6.jpg", "media/0ae1e1413171c6ddb65a/F7.jpg", "media/0ae1e1413171c6ddb65a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f8424c0b08354e78a0af2158ce3016d19c692a05492f95ce15f4393301f02e3"}
{"qa_id": "MVS-Synth__0115__camera_view_direction_change__00", "video_id": "MVS-Synth__0115", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/d669ab385a114e754df9/F0.jpg", "media/d669ab385a114e754df9/F1.jpg", "media/d669ab385a114e754df9/F2.jpg", "media/d669ab385a114e754df9/F3.jpg", "media/d669ab385a114e754df9/F4.jpg", "media/d669ab385a114e754df9/F5.jpg", "media/d669ab385a114e754df9/F6.jpg", "media/d669ab385a114e754df9/F7.jpg", "media/d669ab385a114e754df9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1912eec683eea1dfcf313a947a080c7cba700c52619bb2eddcbd50df4efaec8a"}
{"qa_id": "MVS-Synth__0116__camera_translation_direction__00", "video_id": "MVS-Synth__0116", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "right", "down", "forward"], "frame_paths": ["media/1fc32e6e4d75783a3121/F0.jpg", "media/1fc32e6e4d75783a3121/F1.jpg", "media/1fc32e6e4d75783a3121/F2.jpg", "media/1fc32e6e4d75783a3121/F3.jpg", "media/1fc32e6e4d75783a3121/F4.jpg", "media/1fc32e6e4d75783a3121/F5.jpg", "media/1fc32e6e4d75783a3121/F6.jpg", "media/1fc32e6e4d75783a3121/F7.jpg", "media/1fc32e6e4d75783a3121/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c38dd11d63f137a8dde673263ddf3df83bc4bb83266aed0d14e2d465ca1d3cc"}
{"qa_id": "MVS-Synth__0116__camera_view_direction_change__01", "video_id": "MVS-Synth__0116", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/1fc32e6e4d75783a3121/F0.jpg", "media/1fc32e6e4d75783a3121/F1.jpg", "media/1fc32e6e4d75783a3121/F2.jpg", "media/1fc32e6e4d75783a3121/F3.jpg", "media/1fc32e6e4d75783a3121/F4.jpg", "media/1fc32e6e4d75783a3121/F5.jpg", "media/1fc32e6e4d75783a3121/F6.jpg", "media/1fc32e6e4d75783a3121/F7.jpg", "media/1fc32e6e4d75783a3121/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "MVS-Synth__0117__camera_translation_direction__00", "video_id": "MVS-Synth__0117", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/8ead63891a0d7d79c876/F0.jpg", "media/8ead63891a0d7d79c876/F1.jpg", "media/8ead63891a0d7d79c876/F2.jpg", "media/8ead63891a0d7d79c876/F3.jpg", "media/8ead63891a0d7d79c876/F4.jpg", "media/8ead63891a0d7d79c876/F5.jpg", "media/8ead63891a0d7d79c876/F6.jpg", "media/8ead63891a0d7d79c876/F7.jpg", "media/8ead63891a0d7d79c876/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e99c6910f74fde11d6780335aa77a41aeac22a1cc06ff2194d44a5e468d4eb99"}
{"qa_id": "MVS-Synth__0117__camera_view_direction_change__01", "video_id": "MVS-Synth__0117", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/8ead63891a0d7d79c876/F0.jpg", "media/8ead63891a0d7d79c876/F1.jpg", "media/8ead63891a0d7d79c876/F2.jpg", "media/8ead63891a0d7d79c876/F3.jpg", "media/8ead63891a0d7d79c876/F4.jpg", "media/8ead63891a0d7d79c876/F5.jpg", "media/8ead63891a0d7d79c876/F6.jpg", "media/8ead63891a0d7d79c876/F7.jpg", "media/8ead63891a0d7d79c876/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "70b437908da555037b70ae35a5c890efcf5676cc51943750ef0d29ca9b6b097d"}
{"qa_id": "MVS-Synth__0118__camera_translation_direction__00", "video_id": "MVS-Synth__0118", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/99815373cb5e139bf361/F0.jpg", "media/99815373cb5e139bf361/F1.jpg", "media/99815373cb5e139bf361/F2.jpg", "media/99815373cb5e139bf361/F3.jpg", "media/99815373cb5e139bf361/F4.jpg", "media/99815373cb5e139bf361/F5.jpg", "media/99815373cb5e139bf361/F6.jpg", "media/99815373cb5e139bf361/F7.jpg", "media/99815373cb5e139bf361/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "20391fb8c9a73e5d3198232d46d9e7572b79f47ad743ca8eef6d0044b7f00e0b"}
{"qa_id": "MVS-Synth__0118__camera_view_direction_change__01", "video_id": "MVS-Synth__0118", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/99815373cb5e139bf361/F0.jpg", "media/99815373cb5e139bf361/F1.jpg", "media/99815373cb5e139bf361/F2.jpg", "media/99815373cb5e139bf361/F3.jpg", "media/99815373cb5e139bf361/F4.jpg", "media/99815373cb5e139bf361/F5.jpg", "media/99815373cb5e139bf361/F6.jpg", "media/99815373cb5e139bf361/F7.jpg", "media/99815373cb5e139bf361/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe49dec5b326f5c289d8be859ecbaac9b78bbdb988deb8691948135fcd374071"}
{"qa_id": "MVS-Synth__0119__camera_view_direction_change__00", "video_id": "MVS-Synth__0119", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/4314de0f58a29d11af5a/F0.jpg", "media/4314de0f58a29d11af5a/F1.jpg", "media/4314de0f58a29d11af5a/F2.jpg", "media/4314de0f58a29d11af5a/F3.jpg", "media/4314de0f58a29d11af5a/F4.jpg", "media/4314de0f58a29d11af5a/F5.jpg", "media/4314de0f58a29d11af5a/F6.jpg", "media/4314de0f58a29d11af5a/F7.jpg", "media/4314de0f58a29d11af5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a4ece1b8f8e1b618ba576b991ffabe56bd928190e1bb53ffe2bf5a4c5941650f"}
{"qa_id": "SAV__sav_000144_manual__camera_view_direction_change__00", "video_id": "SAV__sav_000144_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/a9268b26e5cf601632da/F0.jpg", "media/a9268b26e5cf601632da/F1.jpg", "media/a9268b26e5cf601632da/F2.jpg", "media/a9268b26e5cf601632da/F3.jpg", "media/a9268b26e5cf601632da/F4.jpg", "media/a9268b26e5cf601632da/F5.jpg", "media/a9268b26e5cf601632da/F6.jpg", "media/a9268b26e5cf601632da/F7.jpg", "media/a9268b26e5cf601632da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "be2ef52ec9261eabbede0c7cb6166a77dcee89847a47932521436ccc811ff72a"}
{"qa_id": "SAV__sav_000144_manual__object_travel_distance_comparison__01", "video_id": "SAV__sav_000144_manual", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in long black coat with beige shoulder bag) or B (person in black jacket and pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/ca46971f6edc5a0e6ea3/F0.jpg", "media/ca46971f6edc5a0e6ea3/F1.jpg", "media/ca46971f6edc5a0e6ea3/F2.jpg", "media/ca46971f6edc5a0e6ea3/F3.jpg", "media/ca46971f6edc5a0e6ea3/F4.jpg", "media/ca46971f6edc5a0e6ea3/F5.jpg", "media/ca46971f6edc5a0e6ea3/F6.jpg", "media/ca46971f6edc5a0e6ea3/F7.jpg", "media/ca46971f6edc5a0e6ea3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in long black coat with beige shoulder bag) or B (person in black jacket and pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca0abb96b877aa5ac72833df31746b1b960a1828f9e06bfc7d1174186b089dfb"}
{"qa_id": "SAV__sav_000840_manual__camera_view_direction_change__00", "video_id": "SAV__sav_000840_manual", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/a104bb368722856f6910/F0.jpg", "media/a104bb368722856f6910/F1.jpg", "media/a104bb368722856f6910/F2.jpg", "media/a104bb368722856f6910/F3.jpg", "media/a104bb368722856f6910/F4.jpg", "media/a104bb368722856f6910/F5.jpg", "media/a104bb368722856f6910/F6.jpg", "media/a104bb368722856f6910/F7.jpg", "media/a104bb368722856f6910/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "308fd95dc628d4c7c7652aeed45f5498f23bd783280789bdf5dc372effeb021f"}
{"qa_id": "SAV__sav_000840_manual__object_world_motion_direction__01", "video_id": "SAV__sav_000840_manual", "task": "object_world_motion_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in orange top and black pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/b83f27397a5fc5c67734/F0.jpg", "media/b83f27397a5fc5c67734/F1.jpg", "media/b83f27397a5fc5c67734/F2.jpg", "media/b83f27397a5fc5c67734/F3.jpg", "media/b83f27397a5fc5c67734/F4.jpg", "media/b83f27397a5fc5c67734/F5.jpg", "media/b83f27397a5fc5c67734/F6.jpg", "media/b83f27397a5fc5c67734/F7.jpg", "media/b83f27397a5fc5c67734/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in orange top and black pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d478cbf34509e55e54e3f67c04e67acabb521b8265ead7968412354c2a04b316"}
{"qa_id": "SAV__sav_000840_manual__camera_object_distance_change__02", "video_id": "SAV__sav_000840_manual", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, did object A (person in orange top and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b83f27397a5fc5c67734/F0.jpg", "media/b83f27397a5fc5c67734/F1.jpg", "media/b83f27397a5fc5c67734/F2.jpg", "media/b83f27397a5fc5c67734/F3.jpg", "media/b83f27397a5fc5c67734/F4.jpg", "media/b83f27397a5fc5c67734/F5.jpg", "media/b83f27397a5fc5c67734/F6.jpg", "media/b83f27397a5fc5c67734/F7.jpg", "media/b83f27397a5fc5c67734/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (person in orange top and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6d80a7dd03058ee1b7e3acc7f7e78d16654ec6e685e6e071ffae96a805f96f7e"}
{"qa_id": "SAV__sav_001982_manual__camera_object_distance_change__00", "video_id": "SAV__sav_001982_manual", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object A (person in dark blue shirt and orange shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ea58bc4666362d334c8c/F0.jpg", "media/ea58bc4666362d334c8c/F1.jpg", "media/ea58bc4666362d334c8c/F2.jpg", "media/ea58bc4666362d334c8c/F3.jpg", "media/ea58bc4666362d334c8c/F4.jpg", "media/ea58bc4666362d334c8c/F5.jpg", "media/ea58bc4666362d334c8c/F6.jpg", "media/ea58bc4666362d334c8c/F7.jpg", "media/ea58bc4666362d334c8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (person in dark blue shirt and orange shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0b05b6e8a50342826e16acba820d63b8aaf33d31f0e8c6f5cc75f5c4a8816a7"}
{"qa_id": "SAV__sav_002251_manual__camera_translation_direction__00", "video_id": "SAV__sav_002251_manual", "task": "camera_translation_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "backward", "right"], "frame_paths": ["media/08b589feabb2595f342a/F0.jpg", "media/08b589feabb2595f342a/F1.jpg", "media/08b589feabb2595f342a/F2.jpg", "media/08b589feabb2595f342a/F3.jpg", "media/08b589feabb2595f342a/F4.jpg", "media/08b589feabb2595f342a/F5.jpg", "media/08b589feabb2595f342a/F6.jpg", "media/08b589feabb2595f342a/F7.jpg", "media/08b589feabb2595f342a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40c7520720996647d688ea2fc5946e79b2f7bb23e27b8fd0fe75b26ea57eb1ed"}
{"qa_id": "SAV__sav_002251_manual__camera_view_direction_change__01", "video_id": "SAV__sav_002251_manual", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/08b589feabb2595f342a/F0.jpg", "media/08b589feabb2595f342a/F1.jpg", "media/08b589feabb2595f342a/F2.jpg", "media/08b589feabb2595f342a/F3.jpg", "media/08b589feabb2595f342a/F4.jpg", "media/08b589feabb2595f342a/F5.jpg", "media/08b589feabb2595f342a/F6.jpg", "media/08b589feabb2595f342a/F7.jpg", "media/08b589feabb2595f342a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c7245bbffe6b55784e041ff78b31c85d03e49d206ac10aeb37180c9b769afa6b"}
{"qa_id": "SAV__sav_003216_manual__camera_translation_direction__00", "video_id": "SAV__sav_003216_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/77a4e93427208578b3c6/F0.jpg", "media/77a4e93427208578b3c6/F1.jpg", "media/77a4e93427208578b3c6/F2.jpg", "media/77a4e93427208578b3c6/F3.jpg", "media/77a4e93427208578b3c6/F4.jpg", "media/77a4e93427208578b3c6/F5.jpg", "media/77a4e93427208578b3c6/F6.jpg", "media/77a4e93427208578b3c6/F7.jpg", "media/77a4e93427208578b3c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "313a277b0405ac03226161a7f70cc2831e7d199f352f5b1186b82940cd321159"}
{"qa_id": "SAV__sav_007408_manual__camera_view_direction_change__00", "video_id": "SAV__sav_007408_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/e9250ffc89db7094a02e/F0.jpg", "media/e9250ffc89db7094a02e/F1.jpg", "media/e9250ffc89db7094a02e/F2.jpg", "media/e9250ffc89db7094a02e/F3.jpg", "media/e9250ffc89db7094a02e/F4.jpg", "media/e9250ffc89db7094a02e/F5.jpg", "media/e9250ffc89db7094a02e/F6.jpg", "media/e9250ffc89db7094a02e/F7.jpg", "media/e9250ffc89db7094a02e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb887c791883f9fedd5f510ff36793cab164de474b8263a37046b669dc47ce98"}
{"qa_id": "SAV__sav_007408_manual__object_world_motion_direction__01", "video_id": "SAV__sav_007408_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (person riding a motorcycle on a dirt road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/16eee170f3c469c89123/F0.jpg", "media/16eee170f3c469c89123/F1.jpg", "media/16eee170f3c469c89123/F2.jpg", "media/16eee170f3c469c89123/F3.jpg", "media/16eee170f3c469c89123/F4.jpg", "media/16eee170f3c469c89123/F5.jpg", "media/16eee170f3c469c89123/F6.jpg", "media/16eee170f3c469c89123/F7.jpg", "media/16eee170f3c469c89123/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (person riding a motorcycle on a dirt road) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05c2bbcb53f7dee2b8ed71bc8097fd5b1474d1c0b0eb54d472d482bd77297927"}
{"qa_id": "SAV__sav_007408_manual__camera_object_distance_change__02", "video_id": "SAV__sav_007408_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F5, did object A (person riding a motorcycle on a dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/16eee170f3c469c89123/F0.jpg", "media/16eee170f3c469c89123/F1.jpg", "media/16eee170f3c469c89123/F2.jpg", "media/16eee170f3c469c89123/F3.jpg", "media/16eee170f3c469c89123/F4.jpg", "media/16eee170f3c469c89123/F5.jpg", "media/16eee170f3c469c89123/F6.jpg", "media/16eee170f3c469c89123/F7.jpg", "media/16eee170f3c469c89123/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (person riding a motorcycle on a dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dd757788335aa27a163f397e0dc5bd8d8dc01c787c034c5395dda4245e97aa83"}
{"qa_id": "SAV__sav_008221_manual__camera_translation_direction__00", "video_id": "SAV__sav_008221_manual", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/242fa51fc7961db4fdb9/F0.jpg", "media/242fa51fc7961db4fdb9/F1.jpg", "media/242fa51fc7961db4fdb9/F2.jpg", "media/242fa51fc7961db4fdb9/F3.jpg", "media/242fa51fc7961db4fdb9/F4.jpg", "media/242fa51fc7961db4fdb9/F5.jpg", "media/242fa51fc7961db4fdb9/F6.jpg", "media/242fa51fc7961db4fdb9/F7.jpg", "media/242fa51fc7961db4fdb9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca6899de83d460e62b701ccf514db1dc0949a72e4a0e00c0ae4e346edd57c31b"}
{"qa_id": "SAV__sav_008221_manual__camera_view_direction_change__01", "video_id": "SAV__sav_008221_manual", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/242fa51fc7961db4fdb9/F0.jpg", "media/242fa51fc7961db4fdb9/F1.jpg", "media/242fa51fc7961db4fdb9/F2.jpg", "media/242fa51fc7961db4fdb9/F3.jpg", "media/242fa51fc7961db4fdb9/F4.jpg", "media/242fa51fc7961db4fdb9/F5.jpg", "media/242fa51fc7961db4fdb9/F6.jpg", "media/242fa51fc7961db4fdb9/F7.jpg", "media/242fa51fc7961db4fdb9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d557c02bd0fb2911f4418b3b154d82d6dd488baf229fcb20e139f938e1b006d"}
{"qa_id": "SAV__sav_010658_manual__camera_view_direction_change__00", "video_id": "SAV__sav_010658_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/cab79fe8931e113a9ba4/F0.jpg", "media/cab79fe8931e113a9ba4/F1.jpg", "media/cab79fe8931e113a9ba4/F2.jpg", "media/cab79fe8931e113a9ba4/F3.jpg", "media/cab79fe8931e113a9ba4/F4.jpg", "media/cab79fe8931e113a9ba4/F5.jpg", "media/cab79fe8931e113a9ba4/F6.jpg", "media/cab79fe8931e113a9ba4/F7.jpg", "media/cab79fe8931e113a9ba4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "SAV__sav_011193_manual__camera_view_direction_change__00", "video_id": "SAV__sav_011193_manual", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/fb14ed62cd2970ea4c48/F0.jpg", "media/fb14ed62cd2970ea4c48/F1.jpg", "media/fb14ed62cd2970ea4c48/F2.jpg", "media/fb14ed62cd2970ea4c48/F3.jpg", "media/fb14ed62cd2970ea4c48/F4.jpg", "media/fb14ed62cd2970ea4c48/F5.jpg", "media/fb14ed62cd2970ea4c48/F6.jpg", "media/fb14ed62cd2970ea4c48/F7.jpg", "media/fb14ed62cd2970ea4c48/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "SAV__sav_013621_manual__camera_translation_direction__00", "video_id": "SAV__sav_013621_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/784a0ba8614ad139dff0/F0.jpg", "media/784a0ba8614ad139dff0/F1.jpg", "media/784a0ba8614ad139dff0/F2.jpg", "media/784a0ba8614ad139dff0/F3.jpg", "media/784a0ba8614ad139dff0/F4.jpg", "media/784a0ba8614ad139dff0/F5.jpg", "media/784a0ba8614ad139dff0/F6.jpg", "media/784a0ba8614ad139dff0/F7.jpg", "media/784a0ba8614ad139dff0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "830d6613a1e7263c7f5dc111774f13aa73bbb2c16bfcbb0239262fb5fdf9ef09"}
{"qa_id": "SAV__sav_013621_manual__camera_view_direction_change__01", "video_id": "SAV__sav_013621_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/784a0ba8614ad139dff0/F0.jpg", "media/784a0ba8614ad139dff0/F1.jpg", "media/784a0ba8614ad139dff0/F2.jpg", "media/784a0ba8614ad139dff0/F3.jpg", "media/784a0ba8614ad139dff0/F4.jpg", "media/784a0ba8614ad139dff0/F5.jpg", "media/784a0ba8614ad139dff0/F6.jpg", "media/784a0ba8614ad139dff0/F7.jpg", "media/784a0ba8614ad139dff0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "SAV__sav_013621_manual__object_world_motion_direction__02", "video_id": "SAV__sav_013621_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object B (person in tan and black hooded coat) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "up", "down"], "frame_paths": ["media/f47664f5ff7a76fdec4a/F0.jpg", "media/f47664f5ff7a76fdec4a/F1.jpg", "media/f47664f5ff7a76fdec4a/F2.jpg", "media/f47664f5ff7a76fdec4a/F3.jpg", "media/f47664f5ff7a76fdec4a/F4.jpg", "media/f47664f5ff7a76fdec4a/F5.jpg", "media/f47664f5ff7a76fdec4a/F6.jpg", "media/f47664f5ff7a76fdec4a/F7.jpg", "media/f47664f5ff7a76fdec4a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object B (person in tan and black hooded coat) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af4a1e964bcd8f0e82c18c2412f9ed6de24f2ce080948cd28814f06e7b2ae626"}
{"qa_id": "SAV__sav_013621_manual__camera_object_distance_change__03", "video_id": "SAV__sav_013621_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object B (person in tan and black hooded coat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f47664f5ff7a76fdec4a/F0.jpg", "media/f47664f5ff7a76fdec4a/F1.jpg", "media/f47664f5ff7a76fdec4a/F2.jpg", "media/f47664f5ff7a76fdec4a/F3.jpg", "media/f47664f5ff7a76fdec4a/F4.jpg", "media/f47664f5ff7a76fdec4a/F5.jpg", "media/f47664f5ff7a76fdec4a/F6.jpg", "media/f47664f5ff7a76fdec4a/F7.jpg", "media/f47664f5ff7a76fdec4a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object B (person in tan and black hooded coat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4ec6fbfaecbd687a603108b8c185b6b71895bb595a27a00703c3d3e2429ae7a9"}
{"qa_id": "SAV__sav_013621_manual__object_pair_distance_change__04", "video_id": "SAV__sav_013621_manual", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did the 3D distance between object A (person in long black coat and striped beanie) and object B (person in tan and black hooded coat) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/766175ccc9d8b9e21889/F0.jpg", "media/766175ccc9d8b9e21889/F1.jpg", "media/766175ccc9d8b9e21889/F2.jpg", "media/766175ccc9d8b9e21889/F3.jpg", "media/766175ccc9d8b9e21889/F4.jpg", "media/766175ccc9d8b9e21889/F5.jpg", "media/766175ccc9d8b9e21889/F6.jpg", "media/766175ccc9d8b9e21889/F7.jpg", "media/766175ccc9d8b9e21889/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (person in long black coat and striped beanie) and object B (person in tan and black hooded coat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a1f91055bd0bae78d9127dbed7ff59c09c6fd2c9cb7172fd128658c573fff896"}
{"qa_id": "SAV__sav_013621_manual__object_travel_distance_comparison__05", "video_id": "SAV__sav_013621_manual", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in long black coat and striped beanie) or B (person in tan and black hooded coat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/766175ccc9d8b9e21889/F0.jpg", "media/766175ccc9d8b9e21889/F1.jpg", "media/766175ccc9d8b9e21889/F2.jpg", "media/766175ccc9d8b9e21889/F3.jpg", "media/766175ccc9d8b9e21889/F4.jpg", "media/766175ccc9d8b9e21889/F5.jpg", "media/766175ccc9d8b9e21889/F6.jpg", "media/766175ccc9d8b9e21889/F7.jpg", "media/766175ccc9d8b9e21889/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in long black coat and striped beanie) or B (person in tan and black hooded coat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d595f101f338a50902235dfc6c0d68442776a83eb4130e5fa1313e3b4b10be3"}
{"qa_id": "SAV__sav_013650_manual__camera_view_direction_change__00", "video_id": "SAV__sav_013650_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/bae97e97f60ff5b78a56/F0.jpg", "media/bae97e97f60ff5b78a56/F1.jpg", "media/bae97e97f60ff5b78a56/F2.jpg", "media/bae97e97f60ff5b78a56/F3.jpg", "media/bae97e97f60ff5b78a56/F4.jpg", "media/bae97e97f60ff5b78a56/F5.jpg", "media/bae97e97f60ff5b78a56/F6.jpg", "media/bae97e97f60ff5b78a56/F7.jpg", "media/bae97e97f60ff5b78a56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "SAV__sav_013867_manual__object_travel_distance_comparison__00", "video_id": "SAV__sav_013867_manual", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in purple shirt) or B (person in green floral dress)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/2d624d3e144aa11dc250/F0.jpg", "media/2d624d3e144aa11dc250/F1.jpg", "media/2d624d3e144aa11dc250/F2.jpg", "media/2d624d3e144aa11dc250/F3.jpg", "media/2d624d3e144aa11dc250/F4.jpg", "media/2d624d3e144aa11dc250/F5.jpg", "media/2d624d3e144aa11dc250/F6.jpg", "media/2d624d3e144aa11dc250/F7.jpg", "media/2d624d3e144aa11dc250/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in purple shirt) or B (person in green floral dress)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "313a6faafaee28a56348f6b29c8c74711de6d4f5c9236979a737cbdd1afb53fc"}
{"qa_id": "SAV__sav_014713_manual__camera_translation_direction__00", "video_id": "SAV__sav_014713_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/47719f5f611cbedd386f/F0.jpg", "media/47719f5f611cbedd386f/F1.jpg", "media/47719f5f611cbedd386f/F2.jpg", "media/47719f5f611cbedd386f/F3.jpg", "media/47719f5f611cbedd386f/F4.jpg", "media/47719f5f611cbedd386f/F5.jpg", "media/47719f5f611cbedd386f/F6.jpg", "media/47719f5f611cbedd386f/F7.jpg", "media/47719f5f611cbedd386f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "909d7bb2dad0e99a088eebf677b8e14c3bf03fae648219a3db6a1c0921d0835c"}
{"qa_id": "SAV__sav_014713_manual__camera_view_direction_change__01", "video_id": "SAV__sav_014713_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/47719f5f611cbedd386f/F0.jpg", "media/47719f5f611cbedd386f/F1.jpg", "media/47719f5f611cbedd386f/F2.jpg", "media/47719f5f611cbedd386f/F3.jpg", "media/47719f5f611cbedd386f/F4.jpg", "media/47719f5f611cbedd386f/F5.jpg", "media/47719f5f611cbedd386f/F6.jpg", "media/47719f5f611cbedd386f/F7.jpg", "media/47719f5f611cbedd386f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a245e62b618f2ca15b6ff17fcece0172f5241f03da0191cbc429ffde38e21fc8"}
{"qa_id": "SAV__sav_014808_manual__camera_view_direction_change__00", "video_id": "SAV__sav_014808_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/2921d4f42724c9320d6b/F0.jpg", "media/2921d4f42724c9320d6b/F1.jpg", "media/2921d4f42724c9320d6b/F2.jpg", "media/2921d4f42724c9320d6b/F3.jpg", "media/2921d4f42724c9320d6b/F4.jpg", "media/2921d4f42724c9320d6b/F5.jpg", "media/2921d4f42724c9320d6b/F6.jpg", "media/2921d4f42724c9320d6b/F7.jpg", "media/2921d4f42724c9320d6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f3290b3246a69f21031c92bcd8f3354b9d9ed5d12e9057d6a997ac7f477ae73"}
{"qa_id": "SAV__sav_015437_manual__camera_view_direction_change__00", "video_id": "SAV__sav_015437_manual", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/3b467946be5d6e967b92/F0.jpg", "media/3b467946be5d6e967b92/F1.jpg", "media/3b467946be5d6e967b92/F2.jpg", "media/3b467946be5d6e967b92/F3.jpg", "media/3b467946be5d6e967b92/F4.jpg", "media/3b467946be5d6e967b92/F5.jpg", "media/3b467946be5d6e967b92/F6.jpg", "media/3b467946be5d6e967b92/F7.jpg", "media/3b467946be5d6e967b92/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f98011f6f5e3bef7856bffbee3a8c2579537b239284b48df76fd5bbadf4bc3f"}
{"qa_id": "SAV__sav_015437_manual__object_world_motion_direction__01", "video_id": "SAV__sav_015437_manual", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (light brown ox with large horns pulling a cart) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/90baafe2f87913dd4585/F0.jpg", "media/90baafe2f87913dd4585/F1.jpg", "media/90baafe2f87913dd4585/F2.jpg", "media/90baafe2f87913dd4585/F3.jpg", "media/90baafe2f87913dd4585/F4.jpg", "media/90baafe2f87913dd4585/F5.jpg", "media/90baafe2f87913dd4585/F6.jpg", "media/90baafe2f87913dd4585/F7.jpg", "media/90baafe2f87913dd4585/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (light brown ox with large horns pulling a cart) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "24df7c6da5c17d17f75cbef4e961ab946f8dd5cb47cb5317b45c62f7132b4e93"}
{"qa_id": "SAV__sav_015437_manual__camera_object_distance_change__02", "video_id": "SAV__sav_015437_manual", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, did object A (light brown ox with large horns pulling a cart) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/90baafe2f87913dd4585/F0.jpg", "media/90baafe2f87913dd4585/F1.jpg", "media/90baafe2f87913dd4585/F2.jpg", "media/90baafe2f87913dd4585/F3.jpg", "media/90baafe2f87913dd4585/F4.jpg", "media/90baafe2f87913dd4585/F5.jpg", "media/90baafe2f87913dd4585/F6.jpg", "media/90baafe2f87913dd4585/F7.jpg", "media/90baafe2f87913dd4585/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object A (light brown ox with large horns pulling a cart) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e1a23aa10e27c361409c5be2376567b993045b6e47864bad56543a808804f247"}
{"qa_id": "SAV__sav_015437_manual__object_travel_distance_comparison__03", "video_id": "SAV__sav_015437_manual", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown ox with large horns pulling a cart) or B (white and grey cow grazing in the grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/414033c3e52336772331/F0.jpg", "media/414033c3e52336772331/F1.jpg", "media/414033c3e52336772331/F2.jpg", "media/414033c3e52336772331/F3.jpg", "media/414033c3e52336772331/F4.jpg", "media/414033c3e52336772331/F5.jpg", "media/414033c3e52336772331/F6.jpg", "media/414033c3e52336772331/F7.jpg", "media/414033c3e52336772331/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown ox with large horns pulling a cart) or B (white and grey cow grazing in the grass)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88cbe5538b8ebd8e73656d36d288a3ebc112882012805c6c54f37846bb856e90"}
{"qa_id": "SAV__sav_016153_manual__camera_translation_direction__00", "video_id": "SAV__sav_016153_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/d9af886bd8733da3a291/F0.jpg", "media/d9af886bd8733da3a291/F1.jpg", "media/d9af886bd8733da3a291/F2.jpg", "media/d9af886bd8733da3a291/F3.jpg", "media/d9af886bd8733da3a291/F4.jpg", "media/d9af886bd8733da3a291/F5.jpg", "media/d9af886bd8733da3a291/F6.jpg", "media/d9af886bd8733da3a291/F7.jpg", "media/d9af886bd8733da3a291/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a05f21aa39c3f678d164b52f3a6799c2d8620fc3afd8c8a4db292632ba9c7a45"}
{"qa_id": "SAV__sav_016304_manual__object_world_motion_direction__00", "video_id": "SAV__sav_016304_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in white tank top and blue shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/7d20d32cb3829af75eab/F0.jpg", "media/7d20d32cb3829af75eab/F1.jpg", "media/7d20d32cb3829af75eab/F2.jpg", "media/7d20d32cb3829af75eab/F3.jpg", "media/7d20d32cb3829af75eab/F4.jpg", "media/7d20d32cb3829af75eab/F5.jpg", "media/7d20d32cb3829af75eab/F6.jpg", "media/7d20d32cb3829af75eab/F7.jpg", "media/7d20d32cb3829af75eab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in white tank top and blue shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e2f7db804dad1ed3db08801817d8a11ec579c3efa9c54b3d0a2523c395f10656"}
{"qa_id": "SAV__sav_016304_manual__camera_object_distance_change__01", "video_id": "SAV__sav_016304_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object A (person in white tank top and blue shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7d20d32cb3829af75eab/F0.jpg", "media/7d20d32cb3829af75eab/F1.jpg", "media/7d20d32cb3829af75eab/F2.jpg", "media/7d20d32cb3829af75eab/F3.jpg", "media/7d20d32cb3829af75eab/F4.jpg", "media/7d20d32cb3829af75eab/F5.jpg", "media/7d20d32cb3829af75eab/F6.jpg", "media/7d20d32cb3829af75eab/F7.jpg", "media/7d20d32cb3829af75eab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (person in white tank top and blue shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce7d8db2ca6b6ab972b3e2affcf6f19c388092ca7bd6fd36289c5aad86bad57e"}
{"qa_id": "SAV__sav_016304_manual__object_pair_distance_change__02", "video_id": "SAV__sav_016304_manual", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (person in white tank top and blue shorts) and object B (white shirt hanging on a clothesline) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/583e06edb803f48729db/F0.jpg", "media/583e06edb803f48729db/F1.jpg", "media/583e06edb803f48729db/F2.jpg", "media/583e06edb803f48729db/F3.jpg", "media/583e06edb803f48729db/F4.jpg", "media/583e06edb803f48729db/F5.jpg", "media/583e06edb803f48729db/F6.jpg", "media/583e06edb803f48729db/F7.jpg", "media/583e06edb803f48729db/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (person in white tank top and blue shorts) and object B (white shirt hanging on a clothesline) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55702739f6be495e526f353cb90d1b1cff1adcf4f4e1ae092c478ad576be2150"}
{"qa_id": "SAV__sav_016304_manual__object_travel_distance_comparison__03", "video_id": "SAV__sav_016304_manual", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white tank top and blue shorts) or B (white shirt hanging on a clothesline)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/583e06edb803f48729db/F0.jpg", "media/583e06edb803f48729db/F1.jpg", "media/583e06edb803f48729db/F2.jpg", "media/583e06edb803f48729db/F3.jpg", "media/583e06edb803f48729db/F4.jpg", "media/583e06edb803f48729db/F5.jpg", "media/583e06edb803f48729db/F6.jpg", "media/583e06edb803f48729db/F7.jpg", "media/583e06edb803f48729db/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white tank top and blue shorts) or B (white shirt hanging on a clothesline)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbf23655c4ab41263781df5e95f647e415dd3fd9a79b9a6b18092013aec19b1a"}
{"qa_id": "SAV__sav_016310_manual__camera_view_direction_change__00", "video_id": "SAV__sav_016310_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/0d07d2c65db22053f793/F0.jpg", "media/0d07d2c65db22053f793/F1.jpg", "media/0d07d2c65db22053f793/F2.jpg", "media/0d07d2c65db22053f793/F3.jpg", "media/0d07d2c65db22053f793/F4.jpg", "media/0d07d2c65db22053f793/F5.jpg", "media/0d07d2c65db22053f793/F6.jpg", "media/0d07d2c65db22053f793/F7.jpg", "media/0d07d2c65db22053f793/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc01c8f565adc7d70c3dfe8415bb1beb47a7031980d38a04b07a94aa6f06829f"}
{"qa_id": "SAV__sav_016310_manual__object_world_motion_direction__01", "video_id": "SAV__sav_016310_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object A (man in light grey shirt and khaki pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/8a311f7772d9c0bc8097/F0.jpg", "media/8a311f7772d9c0bc8097/F1.jpg", "media/8a311f7772d9c0bc8097/F2.jpg", "media/8a311f7772d9c0bc8097/F3.jpg", "media/8a311f7772d9c0bc8097/F4.jpg", "media/8a311f7772d9c0bc8097/F5.jpg", "media/8a311f7772d9c0bc8097/F6.jpg", "media/8a311f7772d9c0bc8097/F7.jpg", "media/8a311f7772d9c0bc8097/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object A (man in light grey shirt and khaki pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4aed41820d0498bb68359e69d0c4901f3fd2c1ddb168c2a1117acc882308a3d1"}
{"qa_id": "SAV__sav_016310_manual__camera_object_distance_change__02", "video_id": "SAV__sav_016310_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, did object A (man in light grey shirt and khaki pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/8a311f7772d9c0bc8097/F0.jpg", "media/8a311f7772d9c0bc8097/F1.jpg", "media/8a311f7772d9c0bc8097/F2.jpg", "media/8a311f7772d9c0bc8097/F3.jpg", "media/8a311f7772d9c0bc8097/F4.jpg", "media/8a311f7772d9c0bc8097/F5.jpg", "media/8a311f7772d9c0bc8097/F6.jpg", "media/8a311f7772d9c0bc8097/F7.jpg", "media/8a311f7772d9c0bc8097/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did object A (man in light grey shirt and khaki pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "648e209ccae5a5e464d54c2b8f80bacdc7dd77ce0f509d38f7fe8a1d69e64d55"}
{"qa_id": "SAV__sav_016433_manual__camera_translation_direction__00", "video_id": "SAV__sav_016433_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/c51ad8925f84b916e0d8/F0.jpg", "media/c51ad8925f84b916e0d8/F1.jpg", "media/c51ad8925f84b916e0d8/F2.jpg", "media/c51ad8925f84b916e0d8/F3.jpg", "media/c51ad8925f84b916e0d8/F4.jpg", "media/c51ad8925f84b916e0d8/F5.jpg", "media/c51ad8925f84b916e0d8/F6.jpg", "media/c51ad8925f84b916e0d8/F7.jpg", "media/c51ad8925f84b916e0d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b27a7184e03c7e7d56a361acd8e3f5bcb6a5740da8df86416f39e3f0f1299f12"}
{"qa_id": "SAV__sav_016663_manual__camera_object_distance_change__00", "video_id": "SAV__sav_016663_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object A (black circular robot vacuum with a central sensor bump) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/71eb9c385a9a097b3707/F0.jpg", "media/71eb9c385a9a097b3707/F1.jpg", "media/71eb9c385a9a097b3707/F2.jpg", "media/71eb9c385a9a097b3707/F3.jpg", "media/71eb9c385a9a097b3707/F4.jpg", "media/71eb9c385a9a097b3707/F5.jpg", "media/71eb9c385a9a097b3707/F6.jpg", "media/71eb9c385a9a097b3707/F7.jpg", "media/71eb9c385a9a097b3707/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (black circular robot vacuum with a central sensor bump) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c2611c41cbf9d476725fb9c03bab053d7e824b3c2bed7597ebb818da3bc58a6"}
{"qa_id": "SAV__sav_017357_manual__camera_view_direction_change__00", "video_id": "SAV__sav_017357_manual", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/ca5ae82effc518c7ec08/F0.jpg", "media/ca5ae82effc518c7ec08/F1.jpg", "media/ca5ae82effc518c7ec08/F2.jpg", "media/ca5ae82effc518c7ec08/F3.jpg", "media/ca5ae82effc518c7ec08/F4.jpg", "media/ca5ae82effc518c7ec08/F5.jpg", "media/ca5ae82effc518c7ec08/F6.jpg", "media/ca5ae82effc518c7ec08/F7.jpg", "media/ca5ae82effc518c7ec08/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "SAV__sav_017357_manual__object_world_motion_direction__01", "video_id": "SAV__sav_017357_manual", "task": "object_world_motion_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (small black and white dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "right", "left", "forward"], "frame_paths": ["media/d0adffb96fef5eea2f85/F0.jpg", "media/d0adffb96fef5eea2f85/F1.jpg", "media/d0adffb96fef5eea2f85/F2.jpg", "media/d0adffb96fef5eea2f85/F3.jpg", "media/d0adffb96fef5eea2f85/F4.jpg", "media/d0adffb96fef5eea2f85/F5.jpg", "media/d0adffb96fef5eea2f85/F6.jpg", "media/d0adffb96fef5eea2f85/F7.jpg", "media/d0adffb96fef5eea2f85/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (small black and white dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3694c490dc3c138e8b28cea83ef03de69e206b3e8224660ae00526fffb3edf7e"}
{"qa_id": "SAV__sav_017609_manual__object_world_motion_direction__00", "video_id": "SAV__sav_017609_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object A (person walking away on a dirt road) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/fc998aad615688471860/F0.jpg", "media/fc998aad615688471860/F1.jpg", "media/fc998aad615688471860/F2.jpg", "media/fc998aad615688471860/F3.jpg", "media/fc998aad615688471860/F4.jpg", "media/fc998aad615688471860/F5.jpg", "media/fc998aad615688471860/F6.jpg", "media/fc998aad615688471860/F7.jpg", "media/fc998aad615688471860/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object A (person walking away on a dirt road) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "488407156fa91699bf20ec08a9740d7334d75deed57520afcb402064f80bcb64"}
{"qa_id": "SAV__sav_017609_manual__camera_object_distance_change__01", "video_id": "SAV__sav_017609_manual", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F7, did object A (person walking away on a dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fc998aad615688471860/F0.jpg", "media/fc998aad615688471860/F1.jpg", "media/fc998aad615688471860/F2.jpg", "media/fc998aad615688471860/F3.jpg", "media/fc998aad615688471860/F4.jpg", "media/fc998aad615688471860/F5.jpg", "media/fc998aad615688471860/F6.jpg", "media/fc998aad615688471860/F7.jpg", "media/fc998aad615688471860/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (person walking away on a dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "33feb386438678ca96dc6c88436c1593326e75191a4aa6a8004d088fdc25cd4e"}
{"qa_id": "SAV__sav_017665_manual__camera_view_direction_change__00", "video_id": "SAV__sav_017665_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/ca4ec439344f6acf6f6b/F0.jpg", "media/ca4ec439344f6acf6f6b/F1.jpg", "media/ca4ec439344f6acf6f6b/F2.jpg", "media/ca4ec439344f6acf6f6b/F3.jpg", "media/ca4ec439344f6acf6f6b/F4.jpg", "media/ca4ec439344f6acf6f6b/F5.jpg", "media/ca4ec439344f6acf6f6b/F6.jpg", "media/ca4ec439344f6acf6f6b/F7.jpg", "media/ca4ec439344f6acf6f6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ede2c58d326312b562a57303fe447e149486d5ba7ab766e41ab05f1bd05038a9"}
{"qa_id": "SAV__sav_017833_manual__camera_view_direction_change__00", "video_id": "SAV__sav_017833_manual", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/ba18858e0a13223258a7/F0.jpg", "media/ba18858e0a13223258a7/F1.jpg", "media/ba18858e0a13223258a7/F2.jpg", "media/ba18858e0a13223258a7/F3.jpg", "media/ba18858e0a13223258a7/F4.jpg", "media/ba18858e0a13223258a7/F5.jpg", "media/ba18858e0a13223258a7/F6.jpg", "media/ba18858e0a13223258a7/F7.jpg", "media/ba18858e0a13223258a7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13e8a1acb3bede661435ab91735c5cbbfe935d0cd5c1b011dda4a116b80e108c"}
{"qa_id": "SAV__sav_017833_manual__camera_object_distance_change__01", "video_id": "SAV__sav_017833_manual", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object A (person wearing a dark t-shirt riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fd87a07bb0d41ad8673e/F0.jpg", "media/fd87a07bb0d41ad8673e/F1.jpg", "media/fd87a07bb0d41ad8673e/F2.jpg", "media/fd87a07bb0d41ad8673e/F3.jpg", "media/fd87a07bb0d41ad8673e/F4.jpg", "media/fd87a07bb0d41ad8673e/F5.jpg", "media/fd87a07bb0d41ad8673e/F6.jpg", "media/fd87a07bb0d41ad8673e/F7.jpg", "media/fd87a07bb0d41ad8673e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (person wearing a dark t-shirt riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c18613728a312322b6e5866d82467792b7f0ebd0980cb5bdcc20d589386d5f7f"}
{"qa_id": "SAV__sav_017833_manual__camera_object_distance_change__02", "video_id": "SAV__sav_017833_manual", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object B (black motorcycle with a visible rear license plate) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c65a71cebae1ab77967f/F0.jpg", "media/c65a71cebae1ab77967f/F1.jpg", "media/c65a71cebae1ab77967f/F2.jpg", "media/c65a71cebae1ab77967f/F3.jpg", "media/c65a71cebae1ab77967f/F4.jpg", "media/c65a71cebae1ab77967f/F5.jpg", "media/c65a71cebae1ab77967f/F6.jpg", "media/c65a71cebae1ab77967f/F7.jpg", "media/c65a71cebae1ab77967f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (black motorcycle with a visible rear license plate) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1d3a693bfc86c0eb3aaced955d521d15f26f373e62b2e2cd433975291027812e"}
{"qa_id": "SAV__sav_017917_manual__object_world_motion_direction__00", "video_id": "SAV__sav_017917_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (boy in blue plaid shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/77b7ce16013b1250fe7a/F0.jpg", "media/77b7ce16013b1250fe7a/F1.jpg", "media/77b7ce16013b1250fe7a/F2.jpg", "media/77b7ce16013b1250fe7a/F3.jpg", "media/77b7ce16013b1250fe7a/F4.jpg", "media/77b7ce16013b1250fe7a/F5.jpg", "media/77b7ce16013b1250fe7a/F6.jpg", "media/77b7ce16013b1250fe7a/F7.jpg", "media/77b7ce16013b1250fe7a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (boy in blue plaid shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a455b7b5277d2e326d80f8a4eeb4592121e4230b0b46924d28111c9e6820cb6"}
{"qa_id": "SAV__sav_017917_manual__camera_object_distance_change__01", "video_id": "SAV__sav_017917_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object A (boy in blue plaid shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/77b7ce16013b1250fe7a/F0.jpg", "media/77b7ce16013b1250fe7a/F1.jpg", "media/77b7ce16013b1250fe7a/F2.jpg", "media/77b7ce16013b1250fe7a/F3.jpg", "media/77b7ce16013b1250fe7a/F4.jpg", "media/77b7ce16013b1250fe7a/F5.jpg", "media/77b7ce16013b1250fe7a/F6.jpg", "media/77b7ce16013b1250fe7a/F7.jpg", "media/77b7ce16013b1250fe7a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (boy in blue plaid shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "54569587b3393f14b936831f68eafb1733507f9f5185c76c38bc8ef77a36013d"}
{"qa_id": "SAV__sav_017917_manual__object_world_motion_direction__02", "video_id": "SAV__sav_017917_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (metal bowl) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "down", "up", "left"], "frame_paths": ["media/f914e90ed97bffdbd84d/F0.jpg", "media/f914e90ed97bffdbd84d/F1.jpg", "media/f914e90ed97bffdbd84d/F2.jpg", "media/f914e90ed97bffdbd84d/F3.jpg", "media/f914e90ed97bffdbd84d/F4.jpg", "media/f914e90ed97bffdbd84d/F5.jpg", "media/f914e90ed97bffdbd84d/F6.jpg", "media/f914e90ed97bffdbd84d/F7.jpg", "media/f914e90ed97bffdbd84d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (metal bowl) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65b573a0ff052f6121732373f6c3752da350d466ce2407194dc618b82791dbd1"}
{"qa_id": "SAV__sav_017917_manual__camera_object_distance_change__03", "video_id": "SAV__sav_017917_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object B (metal bowl) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/f914e90ed97bffdbd84d/F0.jpg", "media/f914e90ed97bffdbd84d/F1.jpg", "media/f914e90ed97bffdbd84d/F2.jpg", "media/f914e90ed97bffdbd84d/F3.jpg", "media/f914e90ed97bffdbd84d/F4.jpg", "media/f914e90ed97bffdbd84d/F5.jpg", "media/f914e90ed97bffdbd84d/F6.jpg", "media/f914e90ed97bffdbd84d/F7.jpg", "media/f914e90ed97bffdbd84d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object B (metal bowl) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "28365361485e975ed9536e151c692ec0af532c3e60c4942baf2fe9094cfbdd24"}
{"qa_id": "SAV__sav_017974_manual__camera_translation_direction__00", "video_id": "SAV__sav_017974_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "left", "right", "forward"], "frame_paths": ["media/52471c8060c73d08efcd/F0.jpg", "media/52471c8060c73d08efcd/F1.jpg", "media/52471c8060c73d08efcd/F2.jpg", "media/52471c8060c73d08efcd/F3.jpg", "media/52471c8060c73d08efcd/F4.jpg", "media/52471c8060c73d08efcd/F5.jpg", "media/52471c8060c73d08efcd/F6.jpg", "media/52471c8060c73d08efcd/F7.jpg", "media/52471c8060c73d08efcd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. left\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "10fd4fdfa83a92267641e3395e45c1a41bf131809667084a837beedc90084e2e"}
{"qa_id": "SAV__sav_017974_manual__object_world_motion_direction__01", "video_id": "SAV__sav_017974_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in pink outfit) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/c9c01a9f63e060bddfa4/F0.jpg", "media/c9c01a9f63e060bddfa4/F1.jpg", "media/c9c01a9f63e060bddfa4/F2.jpg", "media/c9c01a9f63e060bddfa4/F3.jpg", "media/c9c01a9f63e060bddfa4/F4.jpg", "media/c9c01a9f63e060bddfa4/F5.jpg", "media/c9c01a9f63e060bddfa4/F6.jpg", "media/c9c01a9f63e060bddfa4/F7.jpg", "media/c9c01a9f63e060bddfa4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in pink outfit) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5ffbcf8a1382b8312551bc923122eee2b333699ee9e28c6437d06ae934e16d4a"}
{"qa_id": "SAV__sav_017974_manual__object_world_motion_direction__02", "video_id": "SAV__sav_017974_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in blue shirt and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "forward", "down"], "frame_paths": ["media/99f616d380bb760c32ea/F0.jpg", "media/99f616d380bb760c32ea/F1.jpg", "media/99f616d380bb760c32ea/F2.jpg", "media/99f616d380bb760c32ea/F3.jpg", "media/99f616d380bb760c32ea/F4.jpg", "media/99f616d380bb760c32ea/F5.jpg", "media/99f616d380bb760c32ea/F6.jpg", "media/99f616d380bb760c32ea/F7.jpg", "media/99f616d380bb760c32ea/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in blue shirt and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cfd44674ec19ad91a6c7b69629327a3b1af77486354eea405cffc87bcee4593a"}
{"qa_id": "SAV__sav_018148_manual__camera_view_direction_change__00", "video_id": "SAV__sav_018148_manual", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/207e6a7acc29e59cf65f/F0.jpg", "media/207e6a7acc29e59cf65f/F1.jpg", "media/207e6a7acc29e59cf65f/F2.jpg", "media/207e6a7acc29e59cf65f/F3.jpg", "media/207e6a7acc29e59cf65f/F4.jpg", "media/207e6a7acc29e59cf65f/F5.jpg", "media/207e6a7acc29e59cf65f/F6.jpg", "media/207e6a7acc29e59cf65f/F7.jpg", "media/207e6a7acc29e59cf65f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16a4ca6eeb0d84a0d96a1b095761436c4777b97628640bbab69e354a88eb5d55"}
{"qa_id": "SAV__sav_018304_manual__camera_translation_direction__00", "video_id": "SAV__sav_018304_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/8510807b86aba49757f8/F0.jpg", "media/8510807b86aba49757f8/F1.jpg", "media/8510807b86aba49757f8/F2.jpg", "media/8510807b86aba49757f8/F3.jpg", "media/8510807b86aba49757f8/F4.jpg", "media/8510807b86aba49757f8/F5.jpg", "media/8510807b86aba49757f8/F6.jpg", "media/8510807b86aba49757f8/F7.jpg", "media/8510807b86aba49757f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a3c7ce2c59ca8b65b78a3eb763111d68415b64385f193e17e1a7a2c9d31f392e"}
{"qa_id": "SAV__sav_018431_manual__camera_translation_direction__00", "video_id": "SAV__sav_018431_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/f5b89af7c9eccadb606a/F0.jpg", "media/f5b89af7c9eccadb606a/F1.jpg", "media/f5b89af7c9eccadb606a/F2.jpg", "media/f5b89af7c9eccadb606a/F3.jpg", "media/f5b89af7c9eccadb606a/F4.jpg", "media/f5b89af7c9eccadb606a/F5.jpg", "media/f5b89af7c9eccadb606a/F6.jpg", "media/f5b89af7c9eccadb606a/F7.jpg", "media/f5b89af7c9eccadb606a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d20188092f3205ce31cffca967ccce92df0c41bff422dc1a6f38b544d91af49"}
{"qa_id": "SAV__sav_018550_manual__object_world_motion_direction__00", "video_id": "SAV__sav_018550_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "left", "backward", "forward", "down"], "frame_paths": ["media/ba93fd3eb58ed39463fb/F0.jpg", "media/ba93fd3eb58ed39463fb/F1.jpg", "media/ba93fd3eb58ed39463fb/F2.jpg", "media/ba93fd3eb58ed39463fb/F3.jpg", "media/ba93fd3eb58ed39463fb/F4.jpg", "media/ba93fd3eb58ed39463fb/F5.jpg", "media/ba93fd3eb58ed39463fb/F6.jpg", "media/ba93fd3eb58ed39463fb/F7.jpg", "media/ba93fd3eb58ed39463fb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2438652658d833768499e9acbe4daa670f40ddd941b1d65437ac3be38c484c8b"}
{"qa_id": "SAV__sav_018550_manual__camera_object_distance_change__01", "video_id": "SAV__sav_018550_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (person in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ba93fd3eb58ed39463fb/F0.jpg", "media/ba93fd3eb58ed39463fb/F1.jpg", "media/ba93fd3eb58ed39463fb/F2.jpg", "media/ba93fd3eb58ed39463fb/F3.jpg", "media/ba93fd3eb58ed39463fb/F4.jpg", "media/ba93fd3eb58ed39463fb/F5.jpg", "media/ba93fd3eb58ed39463fb/F6.jpg", "media/ba93fd3eb58ed39463fb/F7.jpg", "media/ba93fd3eb58ed39463fb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aad0f2fef5d4e1065f1386001baa0d2905dd81ebcdfc43088b40b981b6ffe52d"}
{"qa_id": "SAV__sav_018658_manual__camera_object_distance_change__00", "video_id": "SAV__sav_018658_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (small orange balance bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/cfa9c45f026e5b66d030/F0.jpg", "media/cfa9c45f026e5b66d030/F1.jpg", "media/cfa9c45f026e5b66d030/F2.jpg", "media/cfa9c45f026e5b66d030/F3.jpg", "media/cfa9c45f026e5b66d030/F4.jpg", "media/cfa9c45f026e5b66d030/F5.jpg", "media/cfa9c45f026e5b66d030/F6.jpg", "media/cfa9c45f026e5b66d030/F7.jpg", "media/cfa9c45f026e5b66d030/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (small orange balance bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f381f46473087c1bcb14b5b15d3925802100305fd4764f5e2689e3d07a3ff898"}
{"qa_id": "SAV__sav_018658_manual__camera_object_distance_change__01", "video_id": "SAV__sav_018658_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object B (child in white and blue shirt riding a bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c7e057239a178a5421d4/F0.jpg", "media/c7e057239a178a5421d4/F1.jpg", "media/c7e057239a178a5421d4/F2.jpg", "media/c7e057239a178a5421d4/F3.jpg", "media/c7e057239a178a5421d4/F4.jpg", "media/c7e057239a178a5421d4/F5.jpg", "media/c7e057239a178a5421d4/F6.jpg", "media/c7e057239a178a5421d4/F7.jpg", "media/c7e057239a178a5421d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (child in white and blue shirt riding a bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "59110b9cd4d8e3d69fc0dee30a5bd1899497ea7bccfd89c5cce7ccc197c64afb"}
{"qa_id": "SAV__sav_018686_manual__camera_view_direction_change__00", "video_id": "SAV__sav_018686_manual", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/08636bdd6e637638ce44/F0.jpg", "media/08636bdd6e637638ce44/F1.jpg", "media/08636bdd6e637638ce44/F2.jpg", "media/08636bdd6e637638ce44/F3.jpg", "media/08636bdd6e637638ce44/F4.jpg", "media/08636bdd6e637638ce44/F5.jpg", "media/08636bdd6e637638ce44/F6.jpg", "media/08636bdd6e637638ce44/F7.jpg", "media/08636bdd6e637638ce44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ac570cf6b336203bd258e90fd80f7b38dabeda9aba416b5a17cca5c167a9fed2"}
{"qa_id": "SAV__sav_019002_manual__camera_translation_direction__00", "video_id": "SAV__sav_019002_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/651ecf92cd793ea16aa9/F0.jpg", "media/651ecf92cd793ea16aa9/F1.jpg", "media/651ecf92cd793ea16aa9/F2.jpg", "media/651ecf92cd793ea16aa9/F3.jpg", "media/651ecf92cd793ea16aa9/F4.jpg", "media/651ecf92cd793ea16aa9/F5.jpg", "media/651ecf92cd793ea16aa9/F6.jpg", "media/651ecf92cd793ea16aa9/F7.jpg", "media/651ecf92cd793ea16aa9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "693511bc8dee1f2ca00877f272ca1d3672cc99dcd7dc9b1a4e4fa2deb9aa154b"}
{"qa_id": "SAV__sav_019002_manual__camera_view_direction_change__01", "video_id": "SAV__sav_019002_manual", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/651ecf92cd793ea16aa9/F0.jpg", "media/651ecf92cd793ea16aa9/F1.jpg", "media/651ecf92cd793ea16aa9/F2.jpg", "media/651ecf92cd793ea16aa9/F3.jpg", "media/651ecf92cd793ea16aa9/F4.jpg", "media/651ecf92cd793ea16aa9/F5.jpg", "media/651ecf92cd793ea16aa9/F6.jpg", "media/651ecf92cd793ea16aa9/F7.jpg", "media/651ecf92cd793ea16aa9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "SAV__sav_019170_manual__camera_view_direction_change__00", "video_id": "SAV__sav_019170_manual", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/96cb895f73546abd1315/F0.jpg", "media/96cb895f73546abd1315/F1.jpg", "media/96cb895f73546abd1315/F2.jpg", "media/96cb895f73546abd1315/F3.jpg", "media/96cb895f73546abd1315/F4.jpg", "media/96cb895f73546abd1315/F5.jpg", "media/96cb895f73546abd1315/F6.jpg", "media/96cb895f73546abd1315/F7.jpg", "media/96cb895f73546abd1315/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "SAV__sav_019178_manual__camera_translation_direction__00", "video_id": "SAV__sav_019178_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/081f229a1dd260d3fb66/F0.jpg", "media/081f229a1dd260d3fb66/F1.jpg", "media/081f229a1dd260d3fb66/F2.jpg", "media/081f229a1dd260d3fb66/F3.jpg", "media/081f229a1dd260d3fb66/F4.jpg", "media/081f229a1dd260d3fb66/F5.jpg", "media/081f229a1dd260d3fb66/F6.jpg", "media/081f229a1dd260d3fb66/F7.jpg", "media/081f229a1dd260d3fb66/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6dc67d45f1fb12b306f3d07d5424cdfa935f44794adb849f2c65064798657790"}
{"qa_id": "SAV__sav_020074_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020074_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (person swimming in the pool) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/c676163050e34e3b54ff/F0.jpg", "media/c676163050e34e3b54ff/F1.jpg", "media/c676163050e34e3b54ff/F2.jpg", "media/c676163050e34e3b54ff/F3.jpg", "media/c676163050e34e3b54ff/F4.jpg", "media/c676163050e34e3b54ff/F5.jpg", "media/c676163050e34e3b54ff/F6.jpg", "media/c676163050e34e3b54ff/F7.jpg", "media/c676163050e34e3b54ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (person swimming in the pool) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "281a88d70179bfb5bf76f0f081d0c6742c419490a4374fbe6982a8728740957f"}
{"qa_id": "SAV__sav_020074_manual__camera_object_distance_change__01", "video_id": "SAV__sav_020074_manual", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F3, did object B (person swimming in the pool) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c676163050e34e3b54ff/F0.jpg", "media/c676163050e34e3b54ff/F1.jpg", "media/c676163050e34e3b54ff/F2.jpg", "media/c676163050e34e3b54ff/F3.jpg", "media/c676163050e34e3b54ff/F4.jpg", "media/c676163050e34e3b54ff/F5.jpg", "media/c676163050e34e3b54ff/F6.jpg", "media/c676163050e34e3b54ff/F7.jpg", "media/c676163050e34e3b54ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object B (person swimming in the pool) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cae5f3289ca02b4b5a6c8cabc298f25cc54693e04664a4c261f3c7a73e0980d3"}
{"qa_id": "SAV__sav_020074_manual__object_pair_distance_change__02", "video_id": "SAV__sav_020074_manual", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F3, did the 3D distance between object A (person swimming in the pool) and object B (person swimming in the pool) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/59c3af5d42e86db83a5d/F0.jpg", "media/59c3af5d42e86db83a5d/F1.jpg", "media/59c3af5d42e86db83a5d/F2.jpg", "media/59c3af5d42e86db83a5d/F3.jpg", "media/59c3af5d42e86db83a5d/F4.jpg", "media/59c3af5d42e86db83a5d/F5.jpg", "media/59c3af5d42e86db83a5d/F6.jpg", "media/59c3af5d42e86db83a5d/F7.jpg", "media/59c3af5d42e86db83a5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (person swimming in the pool) and object B (person swimming in the pool) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "689259a3eccea0a8c78fa076d768cef1cd2ef55bd04f14c4d1518dbd9267d230"}
{"qa_id": "SAV__sav_020074_manual__object_travel_distance_comparison__03", "video_id": "SAV__sav_020074_manual", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person swimming in the pool) or B (person swimming in the pool)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/59c3af5d42e86db83a5d/F0.jpg", "media/59c3af5d42e86db83a5d/F1.jpg", "media/59c3af5d42e86db83a5d/F2.jpg", "media/59c3af5d42e86db83a5d/F3.jpg", "media/59c3af5d42e86db83a5d/F4.jpg", "media/59c3af5d42e86db83a5d/F5.jpg", "media/59c3af5d42e86db83a5d/F6.jpg", "media/59c3af5d42e86db83a5d/F7.jpg", "media/59c3af5d42e86db83a5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person swimming in the pool) or B (person swimming in the pool)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a83bcc51985fd82c2120c7eae843cb5480e025001f68e5d06555403121224bf8"}
{"qa_id": "SAV__sav_020105_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020105_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow standing in a field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "backward", "right"], "frame_paths": ["media/a188166ca31626d9be20/F0.jpg", "media/a188166ca31626d9be20/F1.jpg", "media/a188166ca31626d9be20/F2.jpg", "media/a188166ca31626d9be20/F3.jpg", "media/a188166ca31626d9be20/F4.jpg", "media/a188166ca31626d9be20/F5.jpg", "media/a188166ca31626d9be20/F6.jpg", "media/a188166ca31626d9be20/F7.jpg", "media/a188166ca31626d9be20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow standing in a field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a7add4e02ae77759f050f9acb4b97330fb644d86781637cb724c0fa30ad68e42"}
{"qa_id": "SAV__sav_020116_manual__camera_translation_direction__00", "video_id": "SAV__sav_020116_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "backward", "right"], "frame_paths": ["media/562345e7c2e04fd8a5b8/F0.jpg", "media/562345e7c2e04fd8a5b8/F1.jpg", "media/562345e7c2e04fd8a5b8/F2.jpg", "media/562345e7c2e04fd8a5b8/F3.jpg", "media/562345e7c2e04fd8a5b8/F4.jpg", "media/562345e7c2e04fd8a5b8/F5.jpg", "media/562345e7c2e04fd8a5b8/F6.jpg", "media/562345e7c2e04fd8a5b8/F7.jpg", "media/562345e7c2e04fd8a5b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b39f7bf4ff31c15333e1307b71ed104b39cb739c107ca6b9710086abcaffe49"}
{"qa_id": "SAV__sav_020384_manual__camera_view_direction_change__00", "video_id": "SAV__sav_020384_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/80eb0fd90fc8b2085694/F0.jpg", "media/80eb0fd90fc8b2085694/F1.jpg", "media/80eb0fd90fc8b2085694/F2.jpg", "media/80eb0fd90fc8b2085694/F3.jpg", "media/80eb0fd90fc8b2085694/F4.jpg", "media/80eb0fd90fc8b2085694/F5.jpg", "media/80eb0fd90fc8b2085694/F6.jpg", "media/80eb0fd90fc8b2085694/F7.jpg", "media/80eb0fd90fc8b2085694/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0c0d4f681ca5c7034fac7f869cc4c4fab1f4e9ecce9c982c1e87cf2b8275754"}
{"qa_id": "SAV__sav_020384_manual__camera_object_distance_change__01", "video_id": "SAV__sav_020384_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (person in orange shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fc55c93d56417aace5f4/F0.jpg", "media/fc55c93d56417aace5f4/F1.jpg", "media/fc55c93d56417aace5f4/F2.jpg", "media/fc55c93d56417aace5f4/F3.jpg", "media/fc55c93d56417aace5f4/F4.jpg", "media/fc55c93d56417aace5f4/F5.jpg", "media/fc55c93d56417aace5f4/F6.jpg", "media/fc55c93d56417aace5f4/F7.jpg", "media/fc55c93d56417aace5f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in orange shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eab306183f67261cdd6c846d85f7df7b409f6e0967f2f3011e4eafb5f734bc84"}
{"qa_id": "SAV__sav_020384_manual__camera_object_distance_change__02", "video_id": "SAV__sav_020384_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object B (person in dark shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ed7fbbeee6d857ee5ed2/F0.jpg", "media/ed7fbbeee6d857ee5ed2/F1.jpg", "media/ed7fbbeee6d857ee5ed2/F2.jpg", "media/ed7fbbeee6d857ee5ed2/F3.jpg", "media/ed7fbbeee6d857ee5ed2/F4.jpg", "media/ed7fbbeee6d857ee5ed2/F5.jpg", "media/ed7fbbeee6d857ee5ed2/F6.jpg", "media/ed7fbbeee6d857ee5ed2/F7.jpg", "media/ed7fbbeee6d857ee5ed2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object B (person in dark shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d05fb2cdbfe2a899f3c3bdb62c68f8533c6567938fd6a8f1b439d6bd49159975"}
{"qa_id": "SAV__sav_020386_manual__camera_object_distance_change__00", "video_id": "SAV__sav_020386_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/396a13b270e1af58ca46/F0.jpg", "media/396a13b270e1af58ca46/F1.jpg", "media/396a13b270e1af58ca46/F2.jpg", "media/396a13b270e1af58ca46/F3.jpg", "media/396a13b270e1af58ca46/F4.jpg", "media/396a13b270e1af58ca46/F5.jpg", "media/396a13b270e1af58ca46/F6.jpg", "media/396a13b270e1af58ca46/F7.jpg", "media/396a13b270e1af58ca46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "575aea88c17edb158729c37a2b425f099d881c6b9334622e686eaad4a0b91c5d"}
{"qa_id": "SAV__sav_020464_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020464_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (person wearing a green bib with number 7) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/49631beae0d8475c8375/F0.jpg", "media/49631beae0d8475c8375/F1.jpg", "media/49631beae0d8475c8375/F2.jpg", "media/49631beae0d8475c8375/F3.jpg", "media/49631beae0d8475c8375/F4.jpg", "media/49631beae0d8475c8375/F5.jpg", "media/49631beae0d8475c8375/F6.jpg", "media/49631beae0d8475c8375/F7.jpg", "media/49631beae0d8475c8375/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (person wearing a green bib with number 7) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e42d655f530b35f33a12c211e72caa2c055249333bb7eb63075ecc7396ead7c"}
{"qa_id": "SAV__sav_020464_manual__camera_object_distance_change__01", "video_id": "SAV__sav_020464_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (person wearing a green bib with number 7) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/49631beae0d8475c8375/F0.jpg", "media/49631beae0d8475c8375/F1.jpg", "media/49631beae0d8475c8375/F2.jpg", "media/49631beae0d8475c8375/F3.jpg", "media/49631beae0d8475c8375/F4.jpg", "media/49631beae0d8475c8375/F5.jpg", "media/49631beae0d8475c8375/F6.jpg", "media/49631beae0d8475c8375/F7.jpg", "media/49631beae0d8475c8375/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (person wearing a green bib with number 7) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96c59180e066f2deef06b4694570c24f8d92077bf69bb0429e4b5ae79dae730c"}
{"qa_id": "SAV__sav_020522_manual__camera_view_direction_change__00", "video_id": "SAV__sav_020522_manual", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/7164ee98cb5994e782fe/F0.jpg", "media/7164ee98cb5994e782fe/F1.jpg", "media/7164ee98cb5994e782fe/F2.jpg", "media/7164ee98cb5994e782fe/F3.jpg", "media/7164ee98cb5994e782fe/F4.jpg", "media/7164ee98cb5994e782fe/F5.jpg", "media/7164ee98cb5994e782fe/F6.jpg", "media/7164ee98cb5994e782fe/F7.jpg", "media/7164ee98cb5994e782fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc01c8f565adc7d70c3dfe8415bb1beb47a7031980d38a04b07a94aa6f06829f"}
{"qa_id": "SAV__sav_020522_manual__camera_object_distance_change__01", "video_id": "SAV__sav_020522_manual", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F4, did object A (brown and white dog) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/3d9882fc35d86702debf/F0.jpg", "media/3d9882fc35d86702debf/F1.jpg", "media/3d9882fc35d86702debf/F2.jpg", "media/3d9882fc35d86702debf/F3.jpg", "media/3d9882fc35d86702debf/F4.jpg", "media/3d9882fc35d86702debf/F5.jpg", "media/3d9882fc35d86702debf/F6.jpg", "media/3d9882fc35d86702debf/F7.jpg", "media/3d9882fc35d86702debf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (brown and white dog) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f34063ec52deb711f2c009514356fd18da98dbcdc16542c41f8a0dc276ec02bb"}
{"qa_id": "SAV__sav_020716_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020716_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (taller child in a red shirt and blue pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/2e10044d7145527d8a66/F0.jpg", "media/2e10044d7145527d8a66/F1.jpg", "media/2e10044d7145527d8a66/F2.jpg", "media/2e10044d7145527d8a66/F3.jpg", "media/2e10044d7145527d8a66/F4.jpg", "media/2e10044d7145527d8a66/F5.jpg", "media/2e10044d7145527d8a66/F6.jpg", "media/2e10044d7145527d8a66/F7.jpg", "media/2e10044d7145527d8a66/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (taller child in a red shirt and blue pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9e6701845b76608abf44cfb564bb49d1119d3167146c3268c7840d6239adcdaa"}
{"qa_id": "SAV__sav_020897_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020897_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (red heavy-duty truck with a covered cargo bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/955cef57a642fb4973da/F0.jpg", "media/955cef57a642fb4973da/F1.jpg", "media/955cef57a642fb4973da/F2.jpg", "media/955cef57a642fb4973da/F3.jpg", "media/955cef57a642fb4973da/F4.jpg", "media/955cef57a642fb4973da/F5.jpg", "media/955cef57a642fb4973da/F6.jpg", "media/955cef57a642fb4973da/F7.jpg", "media/955cef57a642fb4973da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (red heavy-duty truck with a covered cargo bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bcf700648f1eb13195d9a0da387e56ac25d05a955f346d9e830d757acb7467a6"}
{"qa_id": "SAV__sav_020990_manual__object_world_motion_direction__00", "video_id": "SAV__sav_020990_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (child in grey shirt and dark shorts playing with a soccer ball) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/564000156fc6edf46d7c/F0.jpg", "media/564000156fc6edf46d7c/F1.jpg", "media/564000156fc6edf46d7c/F2.jpg", "media/564000156fc6edf46d7c/F3.jpg", "media/564000156fc6edf46d7c/F4.jpg", "media/564000156fc6edf46d7c/F5.jpg", "media/564000156fc6edf46d7c/F6.jpg", "media/564000156fc6edf46d7c/F7.jpg", "media/564000156fc6edf46d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (child in grey shirt and dark shorts playing with a soccer ball) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4687db219fa58aec9265be907db9173e60c2f729205abe50bf530a572d01e52c"}
{"qa_id": "SAV__sav_020990_manual__camera_object_distance_change__01", "video_id": "SAV__sav_020990_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (child in grey shirt and dark shorts playing with a soccer ball) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/564000156fc6edf46d7c/F0.jpg", "media/564000156fc6edf46d7c/F1.jpg", "media/564000156fc6edf46d7c/F2.jpg", "media/564000156fc6edf46d7c/F3.jpg", "media/564000156fc6edf46d7c/F4.jpg", "media/564000156fc6edf46d7c/F5.jpg", "media/564000156fc6edf46d7c/F6.jpg", "media/564000156fc6edf46d7c/F7.jpg", "media/564000156fc6edf46d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (child in grey shirt and dark shorts playing with a soccer ball) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2214e402b2f43db1eb6c96efe2ca22bd164b3dab9ebc6800c72b46b76e142302"}
{"qa_id": "SAV__sav_021050_manual__camera_translation_direction__00", "video_id": "SAV__sav_021050_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "backward", "right"], "frame_paths": ["media/19d41aae54867bc87270/F0.jpg", "media/19d41aae54867bc87270/F1.jpg", "media/19d41aae54867bc87270/F2.jpg", "media/19d41aae54867bc87270/F3.jpg", "media/19d41aae54867bc87270/F4.jpg", "media/19d41aae54867bc87270/F5.jpg", "media/19d41aae54867bc87270/F6.jpg", "media/19d41aae54867bc87270/F7.jpg", "media/19d41aae54867bc87270/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "21aefe913ecc4791c5d41c8469ecfbef0cf784864d8f5051af5f3f03fb8d9330"}
{"qa_id": "SAV__sav_021071_manual__camera_translation_direction__00", "video_id": "SAV__sav_021071_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "forward", "backward", "left"], "frame_paths": ["media/eeaf08425661a9e41212/F0.jpg", "media/eeaf08425661a9e41212/F1.jpg", "media/eeaf08425661a9e41212/F2.jpg", "media/eeaf08425661a9e41212/F3.jpg", "media/eeaf08425661a9e41212/F4.jpg", "media/eeaf08425661a9e41212/F5.jpg", "media/eeaf08425661a9e41212/F6.jpg", "media/eeaf08425661a9e41212/F7.jpg", "media/eeaf08425661a9e41212/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "86306f6410a73a3e4aa14c28ed57b420631bc059e8a38ee91711291b0aa01dd9"}
{"qa_id": "SAV__sav_021071_manual__camera_view_direction_change__01", "video_id": "SAV__sav_021071_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/eeaf08425661a9e41212/F0.jpg", "media/eeaf08425661a9e41212/F1.jpg", "media/eeaf08425661a9e41212/F2.jpg", "media/eeaf08425661a9e41212/F3.jpg", "media/eeaf08425661a9e41212/F4.jpg", "media/eeaf08425661a9e41212/F5.jpg", "media/eeaf08425661a9e41212/F6.jpg", "media/eeaf08425661a9e41212/F7.jpg", "media/eeaf08425661a9e41212/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "SAV__sav_021239_manual__camera_translation_direction__00", "video_id": "SAV__sav_021239_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/bc950c8fe1a19ca5e032/F0.jpg", "media/bc950c8fe1a19ca5e032/F1.jpg", "media/bc950c8fe1a19ca5e032/F2.jpg", "media/bc950c8fe1a19ca5e032/F3.jpg", "media/bc950c8fe1a19ca5e032/F4.jpg", "media/bc950c8fe1a19ca5e032/F5.jpg", "media/bc950c8fe1a19ca5e032/F6.jpg", "media/bc950c8fe1a19ca5e032/F7.jpg", "media/bc950c8fe1a19ca5e032/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ed272225e8142b31eda283252129e352fdbc7b451257c8a4e67ba33951f13f68"}
{"qa_id": "SAV__sav_021239_manual__camera_view_direction_change__01", "video_id": "SAV__sav_021239_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/bc950c8fe1a19ca5e032/F0.jpg", "media/bc950c8fe1a19ca5e032/F1.jpg", "media/bc950c8fe1a19ca5e032/F2.jpg", "media/bc950c8fe1a19ca5e032/F3.jpg", "media/bc950c8fe1a19ca5e032/F4.jpg", "media/bc950c8fe1a19ca5e032/F5.jpg", "media/bc950c8fe1a19ca5e032/F6.jpg", "media/bc950c8fe1a19ca5e032/F7.jpg", "media/bc950c8fe1a19ca5e032/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8de08cb458f39739ff5c1cd280c25d142e1834a2fc0ec2ce54635f2dc27b3fec"}
{"qa_id": "SAV__sav_021361_manual__object_world_motion_direction__00", "video_id": "SAV__sav_021361_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in dark clothing walking away) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/b6500ccb1d9ce1c91b09/F0.jpg", "media/b6500ccb1d9ce1c91b09/F1.jpg", "media/b6500ccb1d9ce1c91b09/F2.jpg", "media/b6500ccb1d9ce1c91b09/F3.jpg", "media/b6500ccb1d9ce1c91b09/F4.jpg", "media/b6500ccb1d9ce1c91b09/F5.jpg", "media/b6500ccb1d9ce1c91b09/F6.jpg", "media/b6500ccb1d9ce1c91b09/F7.jpg", "media/b6500ccb1d9ce1c91b09/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (person in dark clothing walking away) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af01b38ddaeb5ac51d50d5e489ecfc071c8b0d9a7dad44436385aa00cdb6ac6b"}
{"qa_id": "SAV__sav_021361_manual__camera_object_distance_change__01", "video_id": "SAV__sav_021361_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (person in dark clothing walking away) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b6500ccb1d9ce1c91b09/F0.jpg", "media/b6500ccb1d9ce1c91b09/F1.jpg", "media/b6500ccb1d9ce1c91b09/F2.jpg", "media/b6500ccb1d9ce1c91b09/F3.jpg", "media/b6500ccb1d9ce1c91b09/F4.jpg", "media/b6500ccb1d9ce1c91b09/F5.jpg", "media/b6500ccb1d9ce1c91b09/F6.jpg", "media/b6500ccb1d9ce1c91b09/F7.jpg", "media/b6500ccb1d9ce1c91b09/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (person in dark clothing walking away) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "008077b8b860efb3f04ecb94cd30d38f4ff8a6408e63234b7a040c5118df6882"}
{"qa_id": "SAV__sav_021366_manual__camera_translation_direction__00", "video_id": "SAV__sav_021366_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "forward", "down", "backward"], "frame_paths": ["media/574aa43eae00a1404337/F0.jpg", "media/574aa43eae00a1404337/F1.jpg", "media/574aa43eae00a1404337/F2.jpg", "media/574aa43eae00a1404337/F3.jpg", "media/574aa43eae00a1404337/F4.jpg", "media/574aa43eae00a1404337/F5.jpg", "media/574aa43eae00a1404337/F6.jpg", "media/574aa43eae00a1404337/F7.jpg", "media/574aa43eae00a1404337/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8a2fee04ef8ce566c2dbd4cbbfd8463ea4d1cf62c1ebb4760e0dec3c9f66919"}
{"qa_id": "SAV__sav_021393_manual__camera_translation_direction__00", "video_id": "SAV__sav_021393_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "forward", "down", "backward"], "frame_paths": ["media/9fd558f6ce413de84107/F0.jpg", "media/9fd558f6ce413de84107/F1.jpg", "media/9fd558f6ce413de84107/F2.jpg", "media/9fd558f6ce413de84107/F3.jpg", "media/9fd558f6ce413de84107/F4.jpg", "media/9fd558f6ce413de84107/F5.jpg", "media/9fd558f6ce413de84107/F6.jpg", "media/9fd558f6ce413de84107/F7.jpg", "media/9fd558f6ce413de84107/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce385b26f37414987845eea31996110be441cef7b8731ca1886ff1126cbb21cd"}
{"qa_id": "SAV__sav_021874_manual__camera_translation_direction__00", "video_id": "SAV__sav_021874_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "left", "right", "down"], "frame_paths": ["media/baf10ad1daf172af8dcd/F0.jpg", "media/baf10ad1daf172af8dcd/F1.jpg", "media/baf10ad1daf172af8dcd/F2.jpg", "media/baf10ad1daf172af8dcd/F3.jpg", "media/baf10ad1daf172af8dcd/F4.jpg", "media/baf10ad1daf172af8dcd/F5.jpg", "media/baf10ad1daf172af8dcd/F6.jpg", "media/baf10ad1daf172af8dcd/F7.jpg", "media/baf10ad1daf172af8dcd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce3fb1b56f132fb56cf05117a35cb9278aeee61d5ecda570002ebe303784110a"}
{"qa_id": "SAV__sav_021976_manual__camera_view_direction_change__00", "video_id": "SAV__sav_021976_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/ff3034919202f33d90d3/F0.jpg", "media/ff3034919202f33d90d3/F1.jpg", "media/ff3034919202f33d90d3/F2.jpg", "media/ff3034919202f33d90d3/F3.jpg", "media/ff3034919202f33d90d3/F4.jpg", "media/ff3034919202f33d90d3/F5.jpg", "media/ff3034919202f33d90d3/F6.jpg", "media/ff3034919202f33d90d3/F7.jpg", "media/ff3034919202f33d90d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a74873eaad6f5b5bc7fa11ba8a6c28fb60d68336f8fd3430e31ab7a810d2526d"}
{"qa_id": "SAV__sav_021976_manual__camera_object_distance_change__01", "video_id": "SAV__sav_021976_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (tan and white dog with a blue collar) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/00020a52b5dcf25ef6f8/F0.jpg", "media/00020a52b5dcf25ef6f8/F1.jpg", "media/00020a52b5dcf25ef6f8/F2.jpg", "media/00020a52b5dcf25ef6f8/F3.jpg", "media/00020a52b5dcf25ef6f8/F4.jpg", "media/00020a52b5dcf25ef6f8/F5.jpg", "media/00020a52b5dcf25ef6f8/F6.jpg", "media/00020a52b5dcf25ef6f8/F7.jpg", "media/00020a52b5dcf25ef6f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (tan and white dog with a blue collar) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "35475f61d87a1ff59317e76d60e233ec2d67be2ffe11fdce9998887fbe927f76"}
{"qa_id": "SAV__sav_022075_manual__object_world_motion_direction__00", "video_id": "SAV__sav_022075_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person wearing a patterned short-sleeve shirt and blue jeans) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/503e2cae8497c39d8203/F0.jpg", "media/503e2cae8497c39d8203/F1.jpg", "media/503e2cae8497c39d8203/F2.jpg", "media/503e2cae8497c39d8203/F3.jpg", "media/503e2cae8497c39d8203/F4.jpg", "media/503e2cae8497c39d8203/F5.jpg", "media/503e2cae8497c39d8203/F6.jpg", "media/503e2cae8497c39d8203/F7.jpg", "media/503e2cae8497c39d8203/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person wearing a patterned short-sleeve shirt and blue jeans) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7dd7e066f4b55986f6d1e4bc7e6337c118dd7762c51adf25975a2fb7babae519"}
{"qa_id": "SAV__sav_022075_manual__camera_object_distance_change__01", "video_id": "SAV__sav_022075_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object A (person wearing a patterned short-sleeve shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/503e2cae8497c39d8203/F0.jpg", "media/503e2cae8497c39d8203/F1.jpg", "media/503e2cae8497c39d8203/F2.jpg", "media/503e2cae8497c39d8203/F3.jpg", "media/503e2cae8497c39d8203/F4.jpg", "media/503e2cae8497c39d8203/F5.jpg", "media/503e2cae8497c39d8203/F6.jpg", "media/503e2cae8497c39d8203/F7.jpg", "media/503e2cae8497c39d8203/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person wearing a patterned short-sleeve shirt and blue jeans) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c72e1373549f3e830f8345b5da38ec06422d96cf419cdfeef61b70bad0c7fdf5"}
{"qa_id": "SAV__sav_022166_manual__camera_translation_direction__00", "video_id": "SAV__sav_022166_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/fbfb8164ba515493fdaa/F0.jpg", "media/fbfb8164ba515493fdaa/F1.jpg", "media/fbfb8164ba515493fdaa/F2.jpg", "media/fbfb8164ba515493fdaa/F3.jpg", "media/fbfb8164ba515493fdaa/F4.jpg", "media/fbfb8164ba515493fdaa/F5.jpg", "media/fbfb8164ba515493fdaa/F6.jpg", "media/fbfb8164ba515493fdaa/F7.jpg", "media/fbfb8164ba515493fdaa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f296e26145a9189ab1aff554f24159ff203bc05b815cffcfc084d1f692517b1a"}
{"qa_id": "SAV__sav_022166_manual__camera_view_direction_change__01", "video_id": "SAV__sav_022166_manual", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/fbfb8164ba515493fdaa/F0.jpg", "media/fbfb8164ba515493fdaa/F1.jpg", "media/fbfb8164ba515493fdaa/F2.jpg", "media/fbfb8164ba515493fdaa/F3.jpg", "media/fbfb8164ba515493fdaa/F4.jpg", "media/fbfb8164ba515493fdaa/F5.jpg", "media/fbfb8164ba515493fdaa/F6.jpg", "media/fbfb8164ba515493fdaa/F7.jpg", "media/fbfb8164ba515493fdaa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "SAV__sav_022166_manual__object_world_motion_direction__02", "video_id": "SAV__sav_022166_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (man in plaid shirt and jeans) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "up", "left", "backward", "forward", "down"], "frame_paths": ["media/5746aab331f2d5f0e5d7/F0.jpg", "media/5746aab331f2d5f0e5d7/F1.jpg", "media/5746aab331f2d5f0e5d7/F2.jpg", "media/5746aab331f2d5f0e5d7/F3.jpg", "media/5746aab331f2d5f0e5d7/F4.jpg", "media/5746aab331f2d5f0e5d7/F5.jpg", "media/5746aab331f2d5f0e5d7/F6.jpg", "media/5746aab331f2d5f0e5d7/F7.jpg", "media/5746aab331f2d5f0e5d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (man in plaid shirt and jeans) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c808837d06625848088779cce86efe8b2b573d075669aa6814340c20a091b7aa"}
{"qa_id": "SAV__sav_022166_manual__object_travel_distance_comparison__03", "video_id": "SAV__sav_022166_manual", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in plaid shirt and jeans) or B (person in white shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/9febabac113800bc0604/F0.jpg", "media/9febabac113800bc0604/F1.jpg", "media/9febabac113800bc0604/F2.jpg", "media/9febabac113800bc0604/F3.jpg", "media/9febabac113800bc0604/F4.jpg", "media/9febabac113800bc0604/F5.jpg", "media/9febabac113800bc0604/F6.jpg", "media/9febabac113800bc0604/F7.jpg", "media/9febabac113800bc0604/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in plaid shirt and jeans) or B (person in white shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "17fb3c7ee7cd11cff53d377e925347d209d6ff107a02388aef7f5e57b9e59d5e"}
{"qa_id": "SAV__sav_022570_manual__camera_view_direction_change__00", "video_id": "SAV__sav_022570_manual", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/02a29c706b64c6d89629/F0.jpg", "media/02a29c706b64c6d89629/F1.jpg", "media/02a29c706b64c6d89629/F2.jpg", "media/02a29c706b64c6d89629/F3.jpg", "media/02a29c706b64c6d89629/F4.jpg", "media/02a29c706b64c6d89629/F5.jpg", "media/02a29c706b64c6d89629/F6.jpg", "media/02a29c706b64c6d89629/F7.jpg", "media/02a29c706b64c6d89629/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "SAV__sav_022775_manual__camera_translation_direction__00", "video_id": "SAV__sav_022775_manual", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/373e9b27643b45d5cddb/F0.jpg", "media/373e9b27643b45d5cddb/F1.jpg", "media/373e9b27643b45d5cddb/F2.jpg", "media/373e9b27643b45d5cddb/F3.jpg", "media/373e9b27643b45d5cddb/F4.jpg", "media/373e9b27643b45d5cddb/F5.jpg", "media/373e9b27643b45d5cddb/F6.jpg", "media/373e9b27643b45d5cddb/F7.jpg", "media/373e9b27643b45d5cddb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3399f411dbefcd899ab83d617b5b04c5398a61443474c24b7d8e7bf67c51d8a5"}
{"qa_id": "SAV__sav_022775_manual__object_world_motion_direction__01", "video_id": "SAV__sav_022775_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (person wearing a blue hooded jacket riding a motorcycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "right", "left", "forward"], "frame_paths": ["media/508de88ed6aeba910005/F0.jpg", "media/508de88ed6aeba910005/F1.jpg", "media/508de88ed6aeba910005/F2.jpg", "media/508de88ed6aeba910005/F3.jpg", "media/508de88ed6aeba910005/F4.jpg", "media/508de88ed6aeba910005/F5.jpg", "media/508de88ed6aeba910005/F6.jpg", "media/508de88ed6aeba910005/F7.jpg", "media/508de88ed6aeba910005/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (person wearing a blue hooded jacket riding a motorcycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd0cac6b19e759720a753e3d8ab59c8abfa1ef3dde02a6bfe0439153fbe376b5"}
{"qa_id": "SAV__sav_022775_manual__camera_object_distance_change__02", "video_id": "SAV__sav_022775_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (person wearing a blue hooded jacket riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/508de88ed6aeba910005/F0.jpg", "media/508de88ed6aeba910005/F1.jpg", "media/508de88ed6aeba910005/F2.jpg", "media/508de88ed6aeba910005/F3.jpg", "media/508de88ed6aeba910005/F4.jpg", "media/508de88ed6aeba910005/F5.jpg", "media/508de88ed6aeba910005/F6.jpg", "media/508de88ed6aeba910005/F7.jpg", "media/508de88ed6aeba910005/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (person wearing a blue hooded jacket riding a motorcycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b41b92cc4bd69c72eedc12647878949441a1d4bd49f03bb3fa5175ecd1a3594c"}
{"qa_id": "SAV__sav_022775_manual__object_world_motion_direction__03", "video_id": "SAV__sav_022775_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (motorcycle with a visible red taillight and green license plate) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/78c3368f089dd848019b/F0.jpg", "media/78c3368f089dd848019b/F1.jpg", "media/78c3368f089dd848019b/F2.jpg", "media/78c3368f089dd848019b/F3.jpg", "media/78c3368f089dd848019b/F4.jpg", "media/78c3368f089dd848019b/F5.jpg", "media/78c3368f089dd848019b/F6.jpg", "media/78c3368f089dd848019b/F7.jpg", "media/78c3368f089dd848019b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (motorcycle with a visible red taillight and green license plate) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "414213edb868ac97334f96058237d787551e9920faeddd9ae53003d314044ddb"}
{"qa_id": "SAV__sav_022775_manual__camera_object_distance_change__04", "video_id": "SAV__sav_022775_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object B (motorcycle with a visible red taillight and green license plate) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/78c3368f089dd848019b/F0.jpg", "media/78c3368f089dd848019b/F1.jpg", "media/78c3368f089dd848019b/F2.jpg", "media/78c3368f089dd848019b/F3.jpg", "media/78c3368f089dd848019b/F4.jpg", "media/78c3368f089dd848019b/F5.jpg", "media/78c3368f089dd848019b/F6.jpg", "media/78c3368f089dd848019b/F7.jpg", "media/78c3368f089dd848019b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object B (motorcycle with a visible red taillight and green license plate) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "07ac6aa33358d75417556287cc4fb543f38db3f27409d22ae35f70af6565e85d"}
{"qa_id": "SAV__sav_023351_manual__object_world_motion_direction__00", "video_id": "SAV__sav_023351_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (black pickup truck with a metal rack on the bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/c9461d69a4bbd1a84e6a/F0.jpg", "media/c9461d69a4bbd1a84e6a/F1.jpg", "media/c9461d69a4bbd1a84e6a/F2.jpg", "media/c9461d69a4bbd1a84e6a/F3.jpg", "media/c9461d69a4bbd1a84e6a/F4.jpg", "media/c9461d69a4bbd1a84e6a/F5.jpg", "media/c9461d69a4bbd1a84e6a/F6.jpg", "media/c9461d69a4bbd1a84e6a/F7.jpg", "media/c9461d69a4bbd1a84e6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (black pickup truck with a metal rack on the bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "241b6291dfe14ab9be0543ea6673bdaa6aa63d87ef95a29f09a47e2cda83c3a1"}
{"qa_id": "SAV__sav_023351_manual__camera_object_distance_change__01", "video_id": "SAV__sav_023351_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (black pickup truck with a metal rack on the bed) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c9461d69a4bbd1a84e6a/F0.jpg", "media/c9461d69a4bbd1a84e6a/F1.jpg", "media/c9461d69a4bbd1a84e6a/F2.jpg", "media/c9461d69a4bbd1a84e6a/F3.jpg", "media/c9461d69a4bbd1a84e6a/F4.jpg", "media/c9461d69a4bbd1a84e6a/F5.jpg", "media/c9461d69a4bbd1a84e6a/F6.jpg", "media/c9461d69a4bbd1a84e6a/F7.jpg", "media/c9461d69a4bbd1a84e6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (black pickup truck with a metal rack on the bed) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a03a46db4197016a326cd134acc8c6a939af81000a07cb6adab7e9401cba5d2"}
{"qa_id": "SAV__sav_023358_manual__camera_translation_direction__00", "video_id": "SAV__sav_023358_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/6925fa1334fa7562862c/F0.jpg", "media/6925fa1334fa7562862c/F1.jpg", "media/6925fa1334fa7562862c/F2.jpg", "media/6925fa1334fa7562862c/F3.jpg", "media/6925fa1334fa7562862c/F4.jpg", "media/6925fa1334fa7562862c/F5.jpg", "media/6925fa1334fa7562862c/F6.jpg", "media/6925fa1334fa7562862c/F7.jpg", "media/6925fa1334fa7562862c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e177657c2302dc315f108e7f3ab0bb0f43989e14d22ae6d75bea8b0426bd805"}
{"qa_id": "SAV__sav_023358_manual__camera_object_distance_change__01", "video_id": "SAV__sav_023358_manual", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object A (person in dark clothing bending over crates) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ac5825d4542b968f2dad/F0.jpg", "media/ac5825d4542b968f2dad/F1.jpg", "media/ac5825d4542b968f2dad/F2.jpg", "media/ac5825d4542b968f2dad/F3.jpg", "media/ac5825d4542b968f2dad/F4.jpg", "media/ac5825d4542b968f2dad/F5.jpg", "media/ac5825d4542b968f2dad/F6.jpg", "media/ac5825d4542b968f2dad/F7.jpg", "media/ac5825d4542b968f2dad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person in dark clothing bending over crates) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cddba534ed2a295177f5958bf0fc8747915942121ab4d371aa8b3f7ccc3b5202"}
{"qa_id": "SAV__sav_023378_manual__object_world_motion_direction__00", "video_id": "SAV__sav_023378_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (brown cow with a white patch on its rump) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/3e37ff98225534b450c1/F0.jpg", "media/3e37ff98225534b450c1/F1.jpg", "media/3e37ff98225534b450c1/F2.jpg", "media/3e37ff98225534b450c1/F3.jpg", "media/3e37ff98225534b450c1/F4.jpg", "media/3e37ff98225534b450c1/F5.jpg", "media/3e37ff98225534b450c1/F6.jpg", "media/3e37ff98225534b450c1/F7.jpg", "media/3e37ff98225534b450c1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (brown cow with a white patch on its rump) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7df84f7be6c095b5967168d236069bd84b36eb340f2a8d60615e146a91405ab8"}
{"qa_id": "SAV__sav_023378_manual__camera_object_distance_change__01", "video_id": "SAV__sav_023378_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, did object A (brown cow with a white patch on its rump) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/3e37ff98225534b450c1/F0.jpg", "media/3e37ff98225534b450c1/F1.jpg", "media/3e37ff98225534b450c1/F2.jpg", "media/3e37ff98225534b450c1/F3.jpg", "media/3e37ff98225534b450c1/F4.jpg", "media/3e37ff98225534b450c1/F5.jpg", "media/3e37ff98225534b450c1/F6.jpg", "media/3e37ff98225534b450c1/F7.jpg", "media/3e37ff98225534b450c1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (brown cow with a white patch on its rump) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "572d34a0389b1eaf383aa69e5ced9c0a4f92c09d8f78babf67d7fe8a0f90030f"}
{"qa_id": "SAV__sav_023744_manual__camera_translation_direction__00", "video_id": "SAV__sav_023744_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/8d80f1783e07660505c4/F0.jpg", "media/8d80f1783e07660505c4/F1.jpg", "media/8d80f1783e07660505c4/F2.jpg", "media/8d80f1783e07660505c4/F3.jpg", "media/8d80f1783e07660505c4/F4.jpg", "media/8d80f1783e07660505c4/F5.jpg", "media/8d80f1783e07660505c4/F6.jpg", "media/8d80f1783e07660505c4/F7.jpg", "media/8d80f1783e07660505c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9211f64c5746848536c7b1e340501233a6d9cacb024e3c792cd6e4d00a56d4e6"}
{"qa_id": "SAV__sav_023744_manual__camera_view_direction_change__01", "video_id": "SAV__sav_023744_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/8d80f1783e07660505c4/F0.jpg", "media/8d80f1783e07660505c4/F1.jpg", "media/8d80f1783e07660505c4/F2.jpg", "media/8d80f1783e07660505c4/F3.jpg", "media/8d80f1783e07660505c4/F4.jpg", "media/8d80f1783e07660505c4/F5.jpg", "media/8d80f1783e07660505c4/F6.jpg", "media/8d80f1783e07660505c4/F7.jpg", "media/8d80f1783e07660505c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c127b0221ff0db64a03f0005a846519a1901b206f1b78e8706c910fecc450f9a"}
{"qa_id": "SAV__sav_023744_manual__camera_object_distance_change__02", "video_id": "SAV__sav_023744_manual", "task": "camera_object_distance_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, did object A (person in blue shirt and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/962e7ebc2efd7bf09b68/F0.jpg", "media/962e7ebc2efd7bf09b68/F1.jpg", "media/962e7ebc2efd7bf09b68/F2.jpg", "media/962e7ebc2efd7bf09b68/F3.jpg", "media/962e7ebc2efd7bf09b68/F4.jpg", "media/962e7ebc2efd7bf09b68/F5.jpg", "media/962e7ebc2efd7bf09b68/F6.jpg", "media/962e7ebc2efd7bf09b68/F7.jpg", "media/962e7ebc2efd7bf09b68/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (person in blue shirt and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "151f1343f7a95000e70bb872bf15ea11bd52c48e5e4272d7d3f41f5d78113a76"}
{"qa_id": "SAV__sav_023855_manual__camera_translation_direction__00", "video_id": "SAV__sav_023855_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/e34e042bc144a482f07d/F0.jpg", "media/e34e042bc144a482f07d/F1.jpg", "media/e34e042bc144a482f07d/F2.jpg", "media/e34e042bc144a482f07d/F3.jpg", "media/e34e042bc144a482f07d/F4.jpg", "media/e34e042bc144a482f07d/F5.jpg", "media/e34e042bc144a482f07d/F6.jpg", "media/e34e042bc144a482f07d/F7.jpg", "media/e34e042bc144a482f07d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5491d0bdf1a3858b38f83efb1e09c8e5b81275d7b37a9d02b3167e2c481dab21"}
{"qa_id": "SAV__sav_023855_manual__object_world_motion_direction__01", "video_id": "SAV__sav_023855_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (red truck with a green tarp covering the cargo bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "right", "left", "forward"], "frame_paths": ["media/ffcda63fef751e38b035/F0.jpg", "media/ffcda63fef751e38b035/F1.jpg", "media/ffcda63fef751e38b035/F2.jpg", "media/ffcda63fef751e38b035/F3.jpg", "media/ffcda63fef751e38b035/F4.jpg", "media/ffcda63fef751e38b035/F5.jpg", "media/ffcda63fef751e38b035/F6.jpg", "media/ffcda63fef751e38b035/F7.jpg", "media/ffcda63fef751e38b035/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (red truck with a green tarp covering the cargo bed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74e03668c79a2c51aa7384da91059b42286852529548aa191cd89d372c5d8eaf"}
{"qa_id": "SAV__sav_023912_manual__camera_translation_direction__00", "video_id": "SAV__sav_023912_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/31d61f7ea7d22cfe0d68/F0.jpg", "media/31d61f7ea7d22cfe0d68/F1.jpg", "media/31d61f7ea7d22cfe0d68/F2.jpg", "media/31d61f7ea7d22cfe0d68/F3.jpg", "media/31d61f7ea7d22cfe0d68/F4.jpg", "media/31d61f7ea7d22cfe0d68/F5.jpg", "media/31d61f7ea7d22cfe0d68/F6.jpg", "media/31d61f7ea7d22cfe0d68/F7.jpg", "media/31d61f7ea7d22cfe0d68/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62a84cbf6ed43c7d1201e5e925b5a906224f15a2afd1f2f81028911fdbe2e50a"}
{"qa_id": "SAV__sav_024015_manual__camera_translation_direction__00", "video_id": "SAV__sav_024015_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/eab3fed0ee08a8e38ecf/F0.jpg", "media/eab3fed0ee08a8e38ecf/F1.jpg", "media/eab3fed0ee08a8e38ecf/F2.jpg", "media/eab3fed0ee08a8e38ecf/F3.jpg", "media/eab3fed0ee08a8e38ecf/F4.jpg", "media/eab3fed0ee08a8e38ecf/F5.jpg", "media/eab3fed0ee08a8e38ecf/F6.jpg", "media/eab3fed0ee08a8e38ecf/F7.jpg", "media/eab3fed0ee08a8e38ecf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "77d604c1dcfd35034abd93c8d9b800f270fce43484467838945b7691b6de62e6"}
{"qa_id": "SAV__sav_024015_manual__object_world_motion_direction__01", "video_id": "SAV__sav_024015_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (child in beige puffy jacket and dark pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/ae52ce837b778b52ca86/F0.jpg", "media/ae52ce837b778b52ca86/F1.jpg", "media/ae52ce837b778b52ca86/F2.jpg", "media/ae52ce837b778b52ca86/F3.jpg", "media/ae52ce837b778b52ca86/F4.jpg", "media/ae52ce837b778b52ca86/F5.jpg", "media/ae52ce837b778b52ca86/F6.jpg", "media/ae52ce837b778b52ca86/F7.jpg", "media/ae52ce837b778b52ca86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (child in beige puffy jacket and dark pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa2cff0585f50ec288d5ae34f312e04391f1018e00e4c1ed270fd7cc85a5cfe9"}
{"qa_id": "SAV__sav_024015_manual__camera_object_distance_change__02", "video_id": "SAV__sav_024015_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (child in beige puffy jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ae52ce837b778b52ca86/F0.jpg", "media/ae52ce837b778b52ca86/F1.jpg", "media/ae52ce837b778b52ca86/F2.jpg", "media/ae52ce837b778b52ca86/F3.jpg", "media/ae52ce837b778b52ca86/F4.jpg", "media/ae52ce837b778b52ca86/F5.jpg", "media/ae52ce837b778b52ca86/F6.jpg", "media/ae52ce837b778b52ca86/F7.jpg", "media/ae52ce837b778b52ca86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (child in beige puffy jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5f89a8b0e61e574b64626c676a067572558a192cb6c220981d791fce3d24a883"}
{"qa_id": "SAV__sav_024015_manual__object_world_motion_direction__03", "video_id": "SAV__sav_024015_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (person in black jacket walking in the background) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "up", "down"], "frame_paths": ["media/8f60a21c0f3c2e790510/F0.jpg", "media/8f60a21c0f3c2e790510/F1.jpg", "media/8f60a21c0f3c2e790510/F2.jpg", "media/8f60a21c0f3c2e790510/F3.jpg", "media/8f60a21c0f3c2e790510/F4.jpg", "media/8f60a21c0f3c2e790510/F5.jpg", "media/8f60a21c0f3c2e790510/F6.jpg", "media/8f60a21c0f3c2e790510/F7.jpg", "media/8f60a21c0f3c2e790510/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (person in black jacket walking in the background) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c2223711d454e8227dab8029b1c6333bfdc567997f99c21df565d72eacf106ed"}
{"qa_id": "SAV__sav_024015_manual__object_travel_distance_comparison__04", "video_id": "SAV__sav_024015_manual", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in beige puffy jacket and dark pants) or B (person in black jacket walking in the background)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/0d4ff026eda06c1834bd/F0.jpg", "media/0d4ff026eda06c1834bd/F1.jpg", "media/0d4ff026eda06c1834bd/F2.jpg", "media/0d4ff026eda06c1834bd/F3.jpg", "media/0d4ff026eda06c1834bd/F4.jpg", "media/0d4ff026eda06c1834bd/F5.jpg", "media/0d4ff026eda06c1834bd/F6.jpg", "media/0d4ff026eda06c1834bd/F7.jpg", "media/0d4ff026eda06c1834bd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in beige puffy jacket and dark pants) or B (person in black jacket walking in the background)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff9189f6c08434f63956373e1b1faa147cad2d2dcef255e63871aa252bdbe47e"}
{"qa_id": "SAV__sav_024144_manual__camera_translation_direction__00", "video_id": "SAV__sav_024144_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "up", "down"], "frame_paths": ["media/577ff48e4168aa594821/F0.jpg", "media/577ff48e4168aa594821/F1.jpg", "media/577ff48e4168aa594821/F2.jpg", "media/577ff48e4168aa594821/F3.jpg", "media/577ff48e4168aa594821/F4.jpg", "media/577ff48e4168aa594821/F5.jpg", "media/577ff48e4168aa594821/F6.jpg", "media/577ff48e4168aa594821/F7.jpg", "media/577ff48e4168aa594821/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5b95b038586f1b0f64f213a9db14a34cbc1ec9c3b6248ce9bd9a8e77902c429"}
{"qa_id": "SAV__sav_024144_manual__camera_view_direction_change__01", "video_id": "SAV__sav_024144_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/577ff48e4168aa594821/F0.jpg", "media/577ff48e4168aa594821/F1.jpg", "media/577ff48e4168aa594821/F2.jpg", "media/577ff48e4168aa594821/F3.jpg", "media/577ff48e4168aa594821/F4.jpg", "media/577ff48e4168aa594821/F5.jpg", "media/577ff48e4168aa594821/F6.jpg", "media/577ff48e4168aa594821/F7.jpg", "media/577ff48e4168aa594821/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "79a144e54bd2cb97522c8bd091b9cc5e6a32f8092390802b208e732a34f9748a"}
{"qa_id": "SAV__sav_024144_manual__object_world_motion_direction__02", "video_id": "SAV__sav_024144_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (person in black clothing) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/892586ccd67c9732b848/F0.jpg", "media/892586ccd67c9732b848/F1.jpg", "media/892586ccd67c9732b848/F2.jpg", "media/892586ccd67c9732b848/F3.jpg", "media/892586ccd67c9732b848/F4.jpg", "media/892586ccd67c9732b848/F5.jpg", "media/892586ccd67c9732b848/F6.jpg", "media/892586ccd67c9732b848/F7.jpg", "media/892586ccd67c9732b848/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (person in black clothing) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38de6957cd1672a13f241765a242ac771b7d9f245c073734659e71274e76cb95"}
{"qa_id": "SAV__sav_024144_manual__camera_object_distance_change__03", "video_id": "SAV__sav_024144_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object A (person in black clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/892586ccd67c9732b848/F0.jpg", "media/892586ccd67c9732b848/F1.jpg", "media/892586ccd67c9732b848/F2.jpg", "media/892586ccd67c9732b848/F3.jpg", "media/892586ccd67c9732b848/F4.jpg", "media/892586ccd67c9732b848/F5.jpg", "media/892586ccd67c9732b848/F6.jpg", "media/892586ccd67c9732b848/F7.jpg", "media/892586ccd67c9732b848/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (person in black clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1196846c41deb7dacd89e664bee83b33c4d6d836cc5cfa3f6178a703a1812011"}
{"qa_id": "SAV__sav_024262_manual__camera_translation_direction__00", "video_id": "SAV__sav_024262_manual", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/982017c1eadc1cbfdb14/F0.jpg", "media/982017c1eadc1cbfdb14/F1.jpg", "media/982017c1eadc1cbfdb14/F2.jpg", "media/982017c1eadc1cbfdb14/F3.jpg", "media/982017c1eadc1cbfdb14/F4.jpg", "media/982017c1eadc1cbfdb14/F5.jpg", "media/982017c1eadc1cbfdb14/F6.jpg", "media/982017c1eadc1cbfdb14/F7.jpg", "media/982017c1eadc1cbfdb14/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b19619b86c3cfcefe83a09b800851f7b63bea66dc3752ee1be4e049f6681bd5d"}
{"qa_id": "SAV__sav_024262_manual__object_world_motion_direction__01", "video_id": "SAV__sav_024262_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (red and white bumper car with a star decal and the number 91 on the back) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/7acc8786370836561d4c/F0.jpg", "media/7acc8786370836561d4c/F1.jpg", "media/7acc8786370836561d4c/F2.jpg", "media/7acc8786370836561d4c/F3.jpg", "media/7acc8786370836561d4c/F4.jpg", "media/7acc8786370836561d4c/F5.jpg", "media/7acc8786370836561d4c/F6.jpg", "media/7acc8786370836561d4c/F7.jpg", "media/7acc8786370836561d4c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (red and white bumper car with a star decal and the number 91 on the back) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "02964cb5df2301859f614194fbd1309c9e2b7606e99f2b461efe70897b93eedb"}
{"qa_id": "SAV__sav_024262_manual__camera_object_distance_change__02", "video_id": "SAV__sav_024262_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (red and white bumper car with a star decal and the number 91 on the back) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7acc8786370836561d4c/F0.jpg", "media/7acc8786370836561d4c/F1.jpg", "media/7acc8786370836561d4c/F2.jpg", "media/7acc8786370836561d4c/F3.jpg", "media/7acc8786370836561d4c/F4.jpg", "media/7acc8786370836561d4c/F5.jpg", "media/7acc8786370836561d4c/F6.jpg", "media/7acc8786370836561d4c/F7.jpg", "media/7acc8786370836561d4c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (red and white bumper car with a star decal and the number 91 on the back) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a1b52af21e5a3b2b7d59e7349283c4f08f792cbc91d2f1c1d37d73f2e7740f2"}
{"qa_id": "SAV__sav_024344_manual__object_world_motion_direction__00", "video_id": "SAV__sav_024344_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in brown uniform carrying a white bucket on their head) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/9623cf722dad621f090b/F0.jpg", "media/9623cf722dad621f090b/F1.jpg", "media/9623cf722dad621f090b/F2.jpg", "media/9623cf722dad621f090b/F3.jpg", "media/9623cf722dad621f090b/F4.jpg", "media/9623cf722dad621f090b/F5.jpg", "media/9623cf722dad621f090b/F6.jpg", "media/9623cf722dad621f090b/F7.jpg", "media/9623cf722dad621f090b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in brown uniform carrying a white bucket on their head) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5ef915dac8541820909a13dc49eb48cee726ad3198002a7408ef183ada14c040"}
{"qa_id": "SAV__sav_024344_manual__camera_object_distance_change__01", "video_id": "SAV__sav_024344_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (person in brown uniform carrying a white bucket on their head) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9623cf722dad621f090b/F0.jpg", "media/9623cf722dad621f090b/F1.jpg", "media/9623cf722dad621f090b/F2.jpg", "media/9623cf722dad621f090b/F3.jpg", "media/9623cf722dad621f090b/F4.jpg", "media/9623cf722dad621f090b/F5.jpg", "media/9623cf722dad621f090b/F6.jpg", "media/9623cf722dad621f090b/F7.jpg", "media/9623cf722dad621f090b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (person in brown uniform carrying a white bucket on their head) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "698f0d39195cbf90a1f7bc3bb6aabd66a7cefe564aa296ae452509da0feff126"}
{"qa_id": "SAV__sav_024421_manual__object_world_motion_direction__00", "video_id": "SAV__sav_024421_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (person in orange shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "down", "backward", "right"], "frame_paths": ["media/7bfd477a4770d41b8970/F0.jpg", "media/7bfd477a4770d41b8970/F1.jpg", "media/7bfd477a4770d41b8970/F2.jpg", "media/7bfd477a4770d41b8970/F3.jpg", "media/7bfd477a4770d41b8970/F4.jpg", "media/7bfd477a4770d41b8970/F5.jpg", "media/7bfd477a4770d41b8970/F6.jpg", "media/7bfd477a4770d41b8970/F7.jpg", "media/7bfd477a4770d41b8970/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (person in orange shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2f11f3fea7d1f76dbe5410699fd2ceeac4cf28fb30e577b441481dfa7d782a7f"}
{"qa_id": "SAV__sav_024544_manual__camera_translation_direction__00", "video_id": "SAV__sav_024544_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "up", "left", "backward", "forward", "down"], "frame_paths": ["media/ac44eb6a73dcfa848b18/F0.jpg", "media/ac44eb6a73dcfa848b18/F1.jpg", "media/ac44eb6a73dcfa848b18/F2.jpg", "media/ac44eb6a73dcfa848b18/F3.jpg", "media/ac44eb6a73dcfa848b18/F4.jpg", "media/ac44eb6a73dcfa848b18/F5.jpg", "media/ac44eb6a73dcfa848b18/F6.jpg", "media/ac44eb6a73dcfa848b18/F7.jpg", "media/ac44eb6a73dcfa848b18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9bfd8fabe109e5b18fc1e1d0afd64f8b6a5965c5c9dfb59d07efae820cb3fffa"}
{"qa_id": "SAV__sav_024544_manual__camera_view_direction_change__01", "video_id": "SAV__sav_024544_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/ac44eb6a73dcfa848b18/F0.jpg", "media/ac44eb6a73dcfa848b18/F1.jpg", "media/ac44eb6a73dcfa848b18/F2.jpg", "media/ac44eb6a73dcfa848b18/F3.jpg", "media/ac44eb6a73dcfa848b18/F4.jpg", "media/ac44eb6a73dcfa848b18/F5.jpg", "media/ac44eb6a73dcfa848b18/F6.jpg", "media/ac44eb6a73dcfa848b18/F7.jpg", "media/ac44eb6a73dcfa848b18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40e30332d0b0c8e1392da70fdf63b7049b1a3fe6bdd5600bf7786d3116dbae03"}
{"qa_id": "SAV__sav_024544_manual__object_world_motion_direction__02", "video_id": "SAV__sav_024544_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in patterned dress holding a stick) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/fb624dfde8fdd98dfcac/F0.jpg", "media/fb624dfde8fdd98dfcac/F1.jpg", "media/fb624dfde8fdd98dfcac/F2.jpg", "media/fb624dfde8fdd98dfcac/F3.jpg", "media/fb624dfde8fdd98dfcac/F4.jpg", "media/fb624dfde8fdd98dfcac/F5.jpg", "media/fb624dfde8fdd98dfcac/F6.jpg", "media/fb624dfde8fdd98dfcac/F7.jpg", "media/fb624dfde8fdd98dfcac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in patterned dress holding a stick) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0c525e7b25574e3b57ddde9b3be7809d70a09705df30914484c6b39134daac51"}
{"qa_id": "SAV__sav_024544_manual__camera_object_distance_change__03", "video_id": "SAV__sav_024544_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (person in patterned dress holding a stick) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fb624dfde8fdd98dfcac/F0.jpg", "media/fb624dfde8fdd98dfcac/F1.jpg", "media/fb624dfde8fdd98dfcac/F2.jpg", "media/fb624dfde8fdd98dfcac/F3.jpg", "media/fb624dfde8fdd98dfcac/F4.jpg", "media/fb624dfde8fdd98dfcac/F5.jpg", "media/fb624dfde8fdd98dfcac/F6.jpg", "media/fb624dfde8fdd98dfcac/F7.jpg", "media/fb624dfde8fdd98dfcac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in patterned dress holding a stick) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e3e3cf7f6ab199caf29c1ed08242f05d5b30780fa586554498107fdee62cc65f"}
{"qa_id": "SAV__sav_024544_manual__object_world_motion_direction__04", "video_id": "SAV__sav_024544_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in red shirt and dark hat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/397683e778b0f237f4b1/F0.jpg", "media/397683e778b0f237f4b1/F1.jpg", "media/397683e778b0f237f4b1/F2.jpg", "media/397683e778b0f237f4b1/F3.jpg", "media/397683e778b0f237f4b1/F4.jpg", "media/397683e778b0f237f4b1/F5.jpg", "media/397683e778b0f237f4b1/F6.jpg", "media/397683e778b0f237f4b1/F7.jpg", "media/397683e778b0f237f4b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in red shirt and dark hat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1ebd84f94867a0b5d9118a9c7f2dbf0f171ad4f4155c7cc7b8f1c86f10cdfd4"}
{"qa_id": "SAV__sav_024544_manual__camera_object_distance_change__05", "video_id": "SAV__sav_024544_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object B (person in red shirt and dark hat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/397683e778b0f237f4b1/F0.jpg", "media/397683e778b0f237f4b1/F1.jpg", "media/397683e778b0f237f4b1/F2.jpg", "media/397683e778b0f237f4b1/F3.jpg", "media/397683e778b0f237f4b1/F4.jpg", "media/397683e778b0f237f4b1/F5.jpg", "media/397683e778b0f237f4b1/F6.jpg", "media/397683e778b0f237f4b1/F7.jpg", "media/397683e778b0f237f4b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object B (person in red shirt and dark hat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7673183d22c381983b6e8858f12ce902c1cc4f621f2d1f958e31194944976296"}
{"qa_id": "SAV__sav_024785_manual__camera_translation_direction__00", "video_id": "SAV__sav_024785_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/f768e18413989415c6c7/F0.jpg", "media/f768e18413989415c6c7/F1.jpg", "media/f768e18413989415c6c7/F2.jpg", "media/f768e18413989415c6c7/F3.jpg", "media/f768e18413989415c6c7/F4.jpg", "media/f768e18413989415c6c7/F5.jpg", "media/f768e18413989415c6c7/F6.jpg", "media/f768e18413989415c6c7/F7.jpg", "media/f768e18413989415c6c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8b764b6556be1c0d1b1c246c52933353a8ef6a8d01db761d934daa35735d98f8"}
{"qa_id": "SAV__sav_024785_manual__camera_view_direction_change__01", "video_id": "SAV__sav_024785_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/f768e18413989415c6c7/F0.jpg", "media/f768e18413989415c6c7/F1.jpg", "media/f768e18413989415c6c7/F2.jpg", "media/f768e18413989415c6c7/F3.jpg", "media/f768e18413989415c6c7/F4.jpg", "media/f768e18413989415c6c7/F5.jpg", "media/f768e18413989415c6c7/F6.jpg", "media/f768e18413989415c6c7/F7.jpg", "media/f768e18413989415c6c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa06e6069661e9dc2b88fb0b656b0d195fc4b50e527bbe95a905ac5eb368e5ce"}
{"qa_id": "SAV__sav_024785_manual__camera_object_distance_change__02", "video_id": "SAV__sav_024785_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object A (child in yellow shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/679710b88c310e1866e3/F0.jpg", "media/679710b88c310e1866e3/F1.jpg", "media/679710b88c310e1866e3/F2.jpg", "media/679710b88c310e1866e3/F3.jpg", "media/679710b88c310e1866e3/F4.jpg", "media/679710b88c310e1866e3/F5.jpg", "media/679710b88c310e1866e3/F6.jpg", "media/679710b88c310e1866e3/F7.jpg", "media/679710b88c310e1866e3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (child in yellow shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0c719b21cc8b1980e3d5d10af385b55558719a641bd65093b2c48bd2ac0a5bfa"}
{"qa_id": "SAV__sav_024821_manual__camera_translation_direction__00", "video_id": "SAV__sav_024821_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/737e9ff4e5aa0665dd36/F0.jpg", "media/737e9ff4e5aa0665dd36/F1.jpg", "media/737e9ff4e5aa0665dd36/F2.jpg", "media/737e9ff4e5aa0665dd36/F3.jpg", "media/737e9ff4e5aa0665dd36/F4.jpg", "media/737e9ff4e5aa0665dd36/F5.jpg", "media/737e9ff4e5aa0665dd36/F6.jpg", "media/737e9ff4e5aa0665dd36/F7.jpg", "media/737e9ff4e5aa0665dd36/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62a84cbf6ed43c7d1201e5e925b5a906224f15a2afd1f2f81028911fdbe2e50a"}
{"qa_id": "SAV__sav_024821_manual__camera_view_direction_change__01", "video_id": "SAV__sav_024821_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/737e9ff4e5aa0665dd36/F0.jpg", "media/737e9ff4e5aa0665dd36/F1.jpg", "media/737e9ff4e5aa0665dd36/F2.jpg", "media/737e9ff4e5aa0665dd36/F3.jpg", "media/737e9ff4e5aa0665dd36/F4.jpg", "media/737e9ff4e5aa0665dd36/F5.jpg", "media/737e9ff4e5aa0665dd36/F6.jpg", "media/737e9ff4e5aa0665dd36/F7.jpg", "media/737e9ff4e5aa0665dd36/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d37d5fa85d4d9077afe0dd33fc16b0db57e17bc76b3f354ad66514d4eed69074"}
{"qa_id": "SAV__sav_024863_manual__camera_translation_direction__00", "video_id": "SAV__sav_024863_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/4a3d0844f5fedfc0072f/F0.jpg", "media/4a3d0844f5fedfc0072f/F1.jpg", "media/4a3d0844f5fedfc0072f/F2.jpg", "media/4a3d0844f5fedfc0072f/F3.jpg", "media/4a3d0844f5fedfc0072f/F4.jpg", "media/4a3d0844f5fedfc0072f/F5.jpg", "media/4a3d0844f5fedfc0072f/F6.jpg", "media/4a3d0844f5fedfc0072f/F7.jpg", "media/4a3d0844f5fedfc0072f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3399f411dbefcd899ab83d617b5b04c5398a61443474c24b7d8e7bf67c51d8a5"}
{"qa_id": "SAV__sav_024885_manual__object_world_motion_direction__00", "video_id": "SAV__sav_024885_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in grey t-shirt and black pants) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "down", "up", "left"], "frame_paths": ["media/2c8a527c64c6ec66eef3/F0.jpg", "media/2c8a527c64c6ec66eef3/F1.jpg", "media/2c8a527c64c6ec66eef3/F2.jpg", "media/2c8a527c64c6ec66eef3/F3.jpg", "media/2c8a527c64c6ec66eef3/F4.jpg", "media/2c8a527c64c6ec66eef3/F5.jpg", "media/2c8a527c64c6ec66eef3/F6.jpg", "media/2c8a527c64c6ec66eef3/F7.jpg", "media/2c8a527c64c6ec66eef3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in grey t-shirt and black pants) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d7ba4f44d50a8419d8ab7aedeb662fcb62efdb634c3c745cde089e275b674d8d"}
{"qa_id": "SAV__sav_024885_manual__camera_object_distance_change__01", "video_id": "SAV__sav_024885_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (person in grey t-shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2c8a527c64c6ec66eef3/F0.jpg", "media/2c8a527c64c6ec66eef3/F1.jpg", "media/2c8a527c64c6ec66eef3/F2.jpg", "media/2c8a527c64c6ec66eef3/F3.jpg", "media/2c8a527c64c6ec66eef3/F4.jpg", "media/2c8a527c64c6ec66eef3/F5.jpg", "media/2c8a527c64c6ec66eef3/F6.jpg", "media/2c8a527c64c6ec66eef3/F7.jpg", "media/2c8a527c64c6ec66eef3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (person in grey t-shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f31c552c3d20d1c32da34e4c2b24a56c17a58f42ac384ba7dee07f7508333708"}
{"qa_id": "SAV__sav_025233_manual__camera_object_distance_change__00", "video_id": "SAV__sav_025233_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object B (person in a grey shirt and dark pants carrying a yellow sack) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/17eccaa40df1533e87c9/F0.jpg", "media/17eccaa40df1533e87c9/F1.jpg", "media/17eccaa40df1533e87c9/F2.jpg", "media/17eccaa40df1533e87c9/F3.jpg", "media/17eccaa40df1533e87c9/F4.jpg", "media/17eccaa40df1533e87c9/F5.jpg", "media/17eccaa40df1533e87c9/F6.jpg", "media/17eccaa40df1533e87c9/F7.jpg", "media/17eccaa40df1533e87c9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (person in a grey shirt and dark pants carrying a yellow sack) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6fe6e8b8629f6f1eb04960e7799b15de3c3ae208e719a21ae8cffc44a4b50597"}
{"qa_id": "SAV__sav_025233_manual__object_pair_distance_change__01", "video_id": "SAV__sav_025233_manual", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did the 3D distance between object A (person in a red and white checkered shirt working on a black tarp) and object B (person in a grey shirt and dark pants carrying a yellow sack) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2e3be73d5abecc05c676/F0.jpg", "media/2e3be73d5abecc05c676/F1.jpg", "media/2e3be73d5abecc05c676/F2.jpg", "media/2e3be73d5abecc05c676/F3.jpg", "media/2e3be73d5abecc05c676/F4.jpg", "media/2e3be73d5abecc05c676/F5.jpg", "media/2e3be73d5abecc05c676/F6.jpg", "media/2e3be73d5abecc05c676/F7.jpg", "media/2e3be73d5abecc05c676/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (person in a red and white checkered shirt working on a black tarp) and object B (person in a grey shirt and dark pants carrying a yellow sack) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "24d6af3e42c19c56c5dd949cd936cc483098d35a8dad9bf40f896f79ecb837c6"}
{"qa_id": "SAV__sav_025233_manual__object_travel_distance_comparison__02", "video_id": "SAV__sav_025233_manual", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in a red and white checkered shirt working on a black tarp) or B (person in a grey shirt and dark pants carrying a yellow sack)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/2e3be73d5abecc05c676/F0.jpg", "media/2e3be73d5abecc05c676/F1.jpg", "media/2e3be73d5abecc05c676/F2.jpg", "media/2e3be73d5abecc05c676/F3.jpg", "media/2e3be73d5abecc05c676/F4.jpg", "media/2e3be73d5abecc05c676/F5.jpg", "media/2e3be73d5abecc05c676/F6.jpg", "media/2e3be73d5abecc05c676/F7.jpg", "media/2e3be73d5abecc05c676/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in a red and white checkered shirt working on a black tarp) or B (person in a grey shirt and dark pants carrying a yellow sack)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "187718ab68eb29e7ad9b02b8955466fed1cd17b5726717cef9081f75dd4c81ed"}
{"qa_id": "SAV__sav_025300_manual__camera_translation_direction__00", "video_id": "SAV__sav_025300_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "backward", "right"], "frame_paths": ["media/9357504330bdeb57919e/F0.jpg", "media/9357504330bdeb57919e/F1.jpg", "media/9357504330bdeb57919e/F2.jpg", "media/9357504330bdeb57919e/F3.jpg", "media/9357504330bdeb57919e/F4.jpg", "media/9357504330bdeb57919e/F5.jpg", "media/9357504330bdeb57919e/F6.jpg", "media/9357504330bdeb57919e/F7.jpg", "media/9357504330bdeb57919e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9ca55e26db3ae4c78a31d5fab36a2f1bb6fb6be178f4d015d412e896205fd6db"}
{"qa_id": "SAV__sav_025300_manual__camera_view_direction_change__01", "video_id": "SAV__sav_025300_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/9357504330bdeb57919e/F0.jpg", "media/9357504330bdeb57919e/F1.jpg", "media/9357504330bdeb57919e/F2.jpg", "media/9357504330bdeb57919e/F3.jpg", "media/9357504330bdeb57919e/F4.jpg", "media/9357504330bdeb57919e/F5.jpg", "media/9357504330bdeb57919e/F6.jpg", "media/9357504330bdeb57919e/F7.jpg", "media/9357504330bdeb57919e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05a6fc3049aa94eea39440d0f79e8cb1485b883475295ac80978922c69053165"}
{"qa_id": "SAV__sav_025300_manual__object_world_motion_direction__02", "video_id": "SAV__sav_025300_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person riding a scooter wearing an orange top) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/fefb94536ae6aa55cf87/F0.jpg", "media/fefb94536ae6aa55cf87/F1.jpg", "media/fefb94536ae6aa55cf87/F2.jpg", "media/fefb94536ae6aa55cf87/F3.jpg", "media/fefb94536ae6aa55cf87/F4.jpg", "media/fefb94536ae6aa55cf87/F5.jpg", "media/fefb94536ae6aa55cf87/F6.jpg", "media/fefb94536ae6aa55cf87/F7.jpg", "media/fefb94536ae6aa55cf87/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person riding a scooter wearing an orange top) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e7dc7bccd9778647e6fd51c896513a6ff36ac83a089842fdc3f3cc27b4a5e60c"}
{"qa_id": "SAV__sav_025300_manual__camera_object_distance_change__03", "video_id": "SAV__sav_025300_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (person riding a scooter wearing an orange top) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fefb94536ae6aa55cf87/F0.jpg", "media/fefb94536ae6aa55cf87/F1.jpg", "media/fefb94536ae6aa55cf87/F2.jpg", "media/fefb94536ae6aa55cf87/F3.jpg", "media/fefb94536ae6aa55cf87/F4.jpg", "media/fefb94536ae6aa55cf87/F5.jpg", "media/fefb94536ae6aa55cf87/F6.jpg", "media/fefb94536ae6aa55cf87/F7.jpg", "media/fefb94536ae6aa55cf87/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person riding a scooter wearing an orange top) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "483a5c6ebcb3732deafea66be509c38628b4b38115c16529f5aeb85ce2dff9bb"}
{"qa_id": "SAV__sav_025398_manual__camera_translation_direction__00", "video_id": "SAV__sav_025398_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "down", "up", "left"], "frame_paths": ["media/e442a812745b60399c85/F0.jpg", "media/e442a812745b60399c85/F1.jpg", "media/e442a812745b60399c85/F2.jpg", "media/e442a812745b60399c85/F3.jpg", "media/e442a812745b60399c85/F4.jpg", "media/e442a812745b60399c85/F5.jpg", "media/e442a812745b60399c85/F6.jpg", "media/e442a812745b60399c85/F7.jpg", "media/e442a812745b60399c85/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8df00dd1ba74551b73d05eb27cea161b95e0b0a24df7e9eaf082c9d5ee040397"}
{"qa_id": "SAV__sav_025398_manual__object_world_motion_direction__01", "video_id": "SAV__sav_025398_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (maroon truck with 'Falcon' text on the rear) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "left", "right", "forward"], "frame_paths": ["media/c350ca1a88ab6c2e2d1c/F0.jpg", "media/c350ca1a88ab6c2e2d1c/F1.jpg", "media/c350ca1a88ab6c2e2d1c/F2.jpg", "media/c350ca1a88ab6c2e2d1c/F3.jpg", "media/c350ca1a88ab6c2e2d1c/F4.jpg", "media/c350ca1a88ab6c2e2d1c/F5.jpg", "media/c350ca1a88ab6c2e2d1c/F6.jpg", "media/c350ca1a88ab6c2e2d1c/F7.jpg", "media/c350ca1a88ab6c2e2d1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (maroon truck with 'Falcon' text on the rear) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. left\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "038134a5b0bc575cafa36f1e6cb63f1795c2be547551124a43118a13c5e69e7d"}
{"qa_id": "SAV__sav_025398_manual__camera_object_distance_change__02", "video_id": "SAV__sav_025398_manual", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did object A (maroon truck with 'Falcon' text on the rear) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c350ca1a88ab6c2e2d1c/F0.jpg", "media/c350ca1a88ab6c2e2d1c/F1.jpg", "media/c350ca1a88ab6c2e2d1c/F2.jpg", "media/c350ca1a88ab6c2e2d1c/F3.jpg", "media/c350ca1a88ab6c2e2d1c/F4.jpg", "media/c350ca1a88ab6c2e2d1c/F5.jpg", "media/c350ca1a88ab6c2e2d1c/F6.jpg", "media/c350ca1a88ab6c2e2d1c/F7.jpg", "media/c350ca1a88ab6c2e2d1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object A (maroon truck with 'Falcon' text on the rear) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "535b16629d02d6df0206364120b2bd788d98addad77f2eac9da881b3b91dd480"}
{"qa_id": "SAV__sav_025430_manual__camera_translation_direction__00", "video_id": "SAV__sav_025430_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/bfec25b4a86d6572a9d2/F0.jpg", "media/bfec25b4a86d6572a9d2/F1.jpg", "media/bfec25b4a86d6572a9d2/F2.jpg", "media/bfec25b4a86d6572a9d2/F3.jpg", "media/bfec25b4a86d6572a9d2/F4.jpg", "media/bfec25b4a86d6572a9d2/F5.jpg", "media/bfec25b4a86d6572a9d2/F6.jpg", "media/bfec25b4a86d6572a9d2/F7.jpg", "media/bfec25b4a86d6572a9d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "882c47ea9db99fcceef0daf2843d7cabebed89d9636ae6059ab143916cbc98a5"}
{"qa_id": "SAV__sav_025430_manual__object_world_motion_direction__01", "video_id": "SAV__sav_025430_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "up", "down"], "frame_paths": ["media/c57b7df80f6788ae9d8d/F0.jpg", "media/c57b7df80f6788ae9d8d/F1.jpg", "media/c57b7df80f6788ae9d8d/F2.jpg", "media/c57b7df80f6788ae9d8d/F3.jpg", "media/c57b7df80f6788ae9d8d/F4.jpg", "media/c57b7df80f6788ae9d8d/F5.jpg", "media/c57b7df80f6788ae9d8d/F6.jpg", "media/c57b7df80f6788ae9d8d/F7.jpg", "media/c57b7df80f6788ae9d8d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "29c0eed143b944aa61f808912303e232ef0af16b8e2e63f8e9abcf451bb1b059"}
{"qa_id": "SAV__sav_025430_manual__object_world_motion_direction__02", "video_id": "SAV__sav_025430_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object B (person in black shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/2c6f4a839efd451d792d/F0.jpg", "media/2c6f4a839efd451d792d/F1.jpg", "media/2c6f4a839efd451d792d/F2.jpg", "media/2c6f4a839efd451d792d/F3.jpg", "media/2c6f4a839efd451d792d/F4.jpg", "media/2c6f4a839efd451d792d/F5.jpg", "media/2c6f4a839efd451d792d/F6.jpg", "media/2c6f4a839efd451d792d/F7.jpg", "media/2c6f4a839efd451d792d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object B (person in black shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "80d428ebb0a515a765a5a9c03a7e72f21e22df647d618b1ed202ee1dfafa4cab"}
{"qa_id": "SAV__sav_025430_manual__object_pair_distance_change__03", "video_id": "SAV__sav_025430_manual", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, did the 3D distance between object A (person in orange shirt) and object B (person in black shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/663bc70e7104fc8e08ad/F0.jpg", "media/663bc70e7104fc8e08ad/F1.jpg", "media/663bc70e7104fc8e08ad/F2.jpg", "media/663bc70e7104fc8e08ad/F3.jpg", "media/663bc70e7104fc8e08ad/F4.jpg", "media/663bc70e7104fc8e08ad/F5.jpg", "media/663bc70e7104fc8e08ad/F6.jpg", "media/663bc70e7104fc8e08ad/F7.jpg", "media/663bc70e7104fc8e08ad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (person in orange shirt) and object B (person in black shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "403f89be91d0c787fc517d515ddaaf0a034ed14e2398e630eeab84827499fd91"}
{"qa_id": "SAV__sav_025445_manual__object_world_motion_direction__00", "video_id": "SAV__sav_025445_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (toddler in yellow vest and grey pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/e51b335c7e631c64daae/F0.jpg", "media/e51b335c7e631c64daae/F1.jpg", "media/e51b335c7e631c64daae/F2.jpg", "media/e51b335c7e631c64daae/F3.jpg", "media/e51b335c7e631c64daae/F4.jpg", "media/e51b335c7e631c64daae/F5.jpg", "media/e51b335c7e631c64daae/F6.jpg", "media/e51b335c7e631c64daae/F7.jpg", "media/e51b335c7e631c64daae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (toddler in yellow vest and grey pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0da9741baa43fc4d0c73b52c245f3aef9bfda490bdd6d849d3fb2c64773e3e57"}
{"qa_id": "SAV__sav_025459_manual__camera_translation_direction__00", "video_id": "SAV__sav_025459_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/00dc5e3f3fd281aa9119/F0.jpg", "media/00dc5e3f3fd281aa9119/F1.jpg", "media/00dc5e3f3fd281aa9119/F2.jpg", "media/00dc5e3f3fd281aa9119/F3.jpg", "media/00dc5e3f3fd281aa9119/F4.jpg", "media/00dc5e3f3fd281aa9119/F5.jpg", "media/00dc5e3f3fd281aa9119/F6.jpg", "media/00dc5e3f3fd281aa9119/F7.jpg", "media/00dc5e3f3fd281aa9119/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b79e245129baa98a10a9d0abe906095722c8c9aac665085cfe9bcf875b7e75cf"}
{"qa_id": "SAV__sav_025459_manual__object_world_motion_direction__01", "video_id": "SAV__sav_025459_manual", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in white shirt and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/a83aa21b25e3f7519c5f/F0.jpg", "media/a83aa21b25e3f7519c5f/F1.jpg", "media/a83aa21b25e3f7519c5f/F2.jpg", "media/a83aa21b25e3f7519c5f/F3.jpg", "media/a83aa21b25e3f7519c5f/F4.jpg", "media/a83aa21b25e3f7519c5f/F5.jpg", "media/a83aa21b25e3f7519c5f/F6.jpg", "media/a83aa21b25e3f7519c5f/F7.jpg", "media/a83aa21b25e3f7519c5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in white shirt and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "02ec32b7deb5323212fde94459704447b195985020cacd615db4be1bf849f50b"}
{"qa_id": "SAV__sav_025459_manual__camera_object_distance_change__02", "video_id": "SAV__sav_025459_manual", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, did object A (person in white shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a83aa21b25e3f7519c5f/F0.jpg", "media/a83aa21b25e3f7519c5f/F1.jpg", "media/a83aa21b25e3f7519c5f/F2.jpg", "media/a83aa21b25e3f7519c5f/F3.jpg", "media/a83aa21b25e3f7519c5f/F4.jpg", "media/a83aa21b25e3f7519c5f/F5.jpg", "media/a83aa21b25e3f7519c5f/F6.jpg", "media/a83aa21b25e3f7519c5f/F7.jpg", "media/a83aa21b25e3f7519c5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (person in white shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65a33c5fd5ddf4552fe5020efe5a4a88c7a4dfc6fe70ae5cafaa6e0bb8724567"}
{"qa_id": "SAV__sav_025517_manual__camera_object_distance_change__00", "video_id": "SAV__sav_025517_manual", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did object A (person wearing a brown jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/27e11e3315f8aa34f43d/F0.jpg", "media/27e11e3315f8aa34f43d/F1.jpg", "media/27e11e3315f8aa34f43d/F2.jpg", "media/27e11e3315f8aa34f43d/F3.jpg", "media/27e11e3315f8aa34f43d/F4.jpg", "media/27e11e3315f8aa34f43d/F5.jpg", "media/27e11e3315f8aa34f43d/F6.jpg", "media/27e11e3315f8aa34f43d/F7.jpg", "media/27e11e3315f8aa34f43d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person wearing a brown jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d3d665ea42be6f5b5ede7289847b8f943cc77370a4e8b259e97be699167cfc06"}
{"qa_id": "SAV__sav_025517_manual__object_pair_distance_change__01", "video_id": "SAV__sav_025517_manual", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did the 3D distance between object A (person wearing a brown jacket) and object B (person wearing a red shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/db41a3a2811385f26d2d/F0.jpg", "media/db41a3a2811385f26d2d/F1.jpg", "media/db41a3a2811385f26d2d/F2.jpg", "media/db41a3a2811385f26d2d/F3.jpg", "media/db41a3a2811385f26d2d/F4.jpg", "media/db41a3a2811385f26d2d/F5.jpg", "media/db41a3a2811385f26d2d/F6.jpg", "media/db41a3a2811385f26d2d/F7.jpg", "media/db41a3a2811385f26d2d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (person wearing a brown jacket) and object B (person wearing a red shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f102769fad4bc50f03bf24d9af019ee67ff461e5cf103d141385ebb8270a2f27"}
{"qa_id": "SAV__sav_025550_manual__camera_object_distance_change__00", "video_id": "SAV__sav_025550_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (person in black jacket and pants with a grey backpack) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/0d4fe3d6b75e80f5c4b8/F0.jpg", "media/0d4fe3d6b75e80f5c4b8/F1.jpg", "media/0d4fe3d6b75e80f5c4b8/F2.jpg", "media/0d4fe3d6b75e80f5c4b8/F3.jpg", "media/0d4fe3d6b75e80f5c4b8/F4.jpg", "media/0d4fe3d6b75e80f5c4b8/F5.jpg", "media/0d4fe3d6b75e80f5c4b8/F6.jpg", "media/0d4fe3d6b75e80f5c4b8/F7.jpg", "media/0d4fe3d6b75e80f5c4b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in black jacket and pants with a grey backpack) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ed726faf1d5be8717d7cd628193bed7c830edbf13b5da73d864cac47a76488f4"}
{"qa_id": "SAV__sav_025601_manual__camera_translation_direction__00", "video_id": "SAV__sav_025601_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "forward", "down", "backward"], "frame_paths": ["media/7339b3c76b5763087792/F0.jpg", "media/7339b3c76b5763087792/F1.jpg", "media/7339b3c76b5763087792/F2.jpg", "media/7339b3c76b5763087792/F3.jpg", "media/7339b3c76b5763087792/F4.jpg", "media/7339b3c76b5763087792/F5.jpg", "media/7339b3c76b5763087792/F6.jpg", "media/7339b3c76b5763087792/F7.jpg", "media/7339b3c76b5763087792/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "882c47ea9db99fcceef0daf2843d7cabebed89d9636ae6059ab143916cbc98a5"}
{"qa_id": "SAV__sav_025601_manual__camera_object_distance_change__01", "video_id": "SAV__sav_025601_manual", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, did object A (person in plaid shirt and maroon pants sitting on a cello sculpture) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/1c334724fe67b2cd8167/F0.jpg", "media/1c334724fe67b2cd8167/F1.jpg", "media/1c334724fe67b2cd8167/F2.jpg", "media/1c334724fe67b2cd8167/F3.jpg", "media/1c334724fe67b2cd8167/F4.jpg", "media/1c334724fe67b2cd8167/F5.jpg", "media/1c334724fe67b2cd8167/F6.jpg", "media/1c334724fe67b2cd8167/F7.jpg", "media/1c334724fe67b2cd8167/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (person in plaid shirt and maroon pants sitting on a cello sculpture) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7700574cd974e013a3e63ab5a2ff872935cfe875f4c6ee3f29905fc85b895bce"}
{"qa_id": "SAV__sav_025607_manual__object_world_motion_direction__00", "video_id": "SAV__sav_025607_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in white shirt and orange shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/2c9821a7b9fddb845e30/F0.jpg", "media/2c9821a7b9fddb845e30/F1.jpg", "media/2c9821a7b9fddb845e30/F2.jpg", "media/2c9821a7b9fddb845e30/F3.jpg", "media/2c9821a7b9fddb845e30/F4.jpg", "media/2c9821a7b9fddb845e30/F5.jpg", "media/2c9821a7b9fddb845e30/F6.jpg", "media/2c9821a7b9fddb845e30/F7.jpg", "media/2c9821a7b9fddb845e30/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in white shirt and orange shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "80f7f95eed8262619265c376f0ad1b3e9e16480bbaa33b5b04b73cdbb4d75c61"}
{"qa_id": "SAV__sav_025607_manual__camera_object_distance_change__01", "video_id": "SAV__sav_025607_manual", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object A (person in white shirt and orange shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2c9821a7b9fddb845e30/F0.jpg", "media/2c9821a7b9fddb845e30/F1.jpg", "media/2c9821a7b9fddb845e30/F2.jpg", "media/2c9821a7b9fddb845e30/F3.jpg", "media/2c9821a7b9fddb845e30/F4.jpg", "media/2c9821a7b9fddb845e30/F5.jpg", "media/2c9821a7b9fddb845e30/F6.jpg", "media/2c9821a7b9fddb845e30/F7.jpg", "media/2c9821a7b9fddb845e30/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person in white shirt and orange shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "26886296c816d040145ffde157acc7df8a97e60189a72265a37dd04b2986c1d8"}
{"qa_id": "SAV__sav_025659_manual__camera_view_direction_change__00", "video_id": "SAV__sav_025659_manual", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/b880549d8973c2537540/F0.jpg", "media/b880549d8973c2537540/F1.jpg", "media/b880549d8973c2537540/F2.jpg", "media/b880549d8973c2537540/F3.jpg", "media/b880549d8973c2537540/F4.jpg", "media/b880549d8973c2537540/F5.jpg", "media/b880549d8973c2537540/F6.jpg", "media/b880549d8973c2537540/F7.jpg", "media/b880549d8973c2537540/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5b0cc595dde2baef28ef1e93927f392683075a5e0cb1d51f67d1de957f260de"}
{"qa_id": "SAV__sav_025659_manual__object_world_motion_direction__01", "video_id": "SAV__sav_025659_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (small child in orange pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "left", "right", "forward"], "frame_paths": ["media/611be95299305da4fac1/F0.jpg", "media/611be95299305da4fac1/F1.jpg", "media/611be95299305da4fac1/F2.jpg", "media/611be95299305da4fac1/F3.jpg", "media/611be95299305da4fac1/F4.jpg", "media/611be95299305da4fac1/F5.jpg", "media/611be95299305da4fac1/F6.jpg", "media/611be95299305da4fac1/F7.jpg", "media/611be95299305da4fac1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (small child in orange pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. left\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5c38489e197bf3a32cef36b9ba744c1642562c81fc21d741d1a747951894b4f"}
{"qa_id": "SAV__sav_025659_manual__object_world_motion_direction__02", "video_id": "SAV__sav_025659_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object B (taller person in grey shirt and dark pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/73bf3f7c79206b77c96c/F0.jpg", "media/73bf3f7c79206b77c96c/F1.jpg", "media/73bf3f7c79206b77c96c/F2.jpg", "media/73bf3f7c79206b77c96c/F3.jpg", "media/73bf3f7c79206b77c96c/F4.jpg", "media/73bf3f7c79206b77c96c/F5.jpg", "media/73bf3f7c79206b77c96c/F6.jpg", "media/73bf3f7c79206b77c96c/F7.jpg", "media/73bf3f7c79206b77c96c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object B (taller person in grey shirt and dark pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2929720b86fe3b8d02d0682c3b4659159c26893860e2a4f72e5ddba5e1534682"}
{"qa_id": "SAV__sav_025862_manual__object_world_motion_direction__00", "video_id": "SAV__sav_025862_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in dark blue shirt and pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/7067ad2aea61e1eacfc5/F0.jpg", "media/7067ad2aea61e1eacfc5/F1.jpg", "media/7067ad2aea61e1eacfc5/F2.jpg", "media/7067ad2aea61e1eacfc5/F3.jpg", "media/7067ad2aea61e1eacfc5/F4.jpg", "media/7067ad2aea61e1eacfc5/F5.jpg", "media/7067ad2aea61e1eacfc5/F6.jpg", "media/7067ad2aea61e1eacfc5/F7.jpg", "media/7067ad2aea61e1eacfc5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in dark blue shirt and pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d77ea013157d49e03eec15ac45e7a955de7c89eefc928d7a073949d6dbcd233"}
{"qa_id": "SAV__sav_025862_manual__object_pair_distance_change__01", "video_id": "SAV__sav_025862_manual", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (person in dark blue shirt and pants) and object B (person in yellow shirt riding a bicycle) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/8f02352eb2013249fa0a/F0.jpg", "media/8f02352eb2013249fa0a/F1.jpg", "media/8f02352eb2013249fa0a/F2.jpg", "media/8f02352eb2013249fa0a/F3.jpg", "media/8f02352eb2013249fa0a/F4.jpg", "media/8f02352eb2013249fa0a/F5.jpg", "media/8f02352eb2013249fa0a/F6.jpg", "media/8f02352eb2013249fa0a/F7.jpg", "media/8f02352eb2013249fa0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (person in dark blue shirt and pants) and object B (person in yellow shirt riding a bicycle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f61353cce3d496ab448d3572c3630bf2e8a2a0c241115078e1e8329b90d84e8"}
{"qa_id": "SAV__sav_025862_manual__object_travel_distance_comparison__02", "video_id": "SAV__sav_025862_manual", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in dark blue shirt and pants) or B (person in yellow shirt riding a bicycle)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/8f02352eb2013249fa0a/F0.jpg", "media/8f02352eb2013249fa0a/F1.jpg", "media/8f02352eb2013249fa0a/F2.jpg", "media/8f02352eb2013249fa0a/F3.jpg", "media/8f02352eb2013249fa0a/F4.jpg", "media/8f02352eb2013249fa0a/F5.jpg", "media/8f02352eb2013249fa0a/F6.jpg", "media/8f02352eb2013249fa0a/F7.jpg", "media/8f02352eb2013249fa0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in dark blue shirt and pants) or B (person in yellow shirt riding a bicycle)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c7da0f0467cc9efd40d8632bbbae76af0cba07c3e4a396e4d14a900844d5d880"}
{"qa_id": "SAV__sav_026730_manual__object_world_motion_direction__00", "video_id": "SAV__sav_026730_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (yellow backhoe loader) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "forward", "down", "backward"], "frame_paths": ["media/c1b3e77c298a827f4077/F0.jpg", "media/c1b3e77c298a827f4077/F1.jpg", "media/c1b3e77c298a827f4077/F2.jpg", "media/c1b3e77c298a827f4077/F3.jpg", "media/c1b3e77c298a827f4077/F4.jpg", "media/c1b3e77c298a827f4077/F5.jpg", "media/c1b3e77c298a827f4077/F6.jpg", "media/c1b3e77c298a827f4077/F7.jpg", "media/c1b3e77c298a827f4077/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (yellow backhoe loader) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "703849237df6c128ac1a6253c7048fa422e3f490c0d911692eb9f4d9b60c31cb"}
{"qa_id": "SAV__sav_026780_manual__camera_translation_direction__00", "video_id": "SAV__sav_026780_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/6937b77af3ac9e96246f/F0.jpg", "media/6937b77af3ac9e96246f/F1.jpg", "media/6937b77af3ac9e96246f/F2.jpg", "media/6937b77af3ac9e96246f/F3.jpg", "media/6937b77af3ac9e96246f/F4.jpg", "media/6937b77af3ac9e96246f/F5.jpg", "media/6937b77af3ac9e96246f/F6.jpg", "media/6937b77af3ac9e96246f/F7.jpg", "media/6937b77af3ac9e96246f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5b0d590b18ec624ee3f915a6fddb7f638dfe471e207bd52d81d9063ce3993e8a"}
{"qa_id": "SAV__sav_026796_manual__camera_translation_direction__00", "video_id": "SAV__sav_026796_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "forward", "down"], "frame_paths": ["media/886b3368c4b680bedf70/F0.jpg", "media/886b3368c4b680bedf70/F1.jpg", "media/886b3368c4b680bedf70/F2.jpg", "media/886b3368c4b680bedf70/F3.jpg", "media/886b3368c4b680bedf70/F4.jpg", "media/886b3368c4b680bedf70/F5.jpg", "media/886b3368c4b680bedf70/F6.jpg", "media/886b3368c4b680bedf70/F7.jpg", "media/886b3368c4b680bedf70/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f8e20b648391e26c9c30f900083beb8364f406c7af6328957a9d8d98be60959"}
{"qa_id": "SAV__sav_026796_manual__camera_view_direction_change__01", "video_id": "SAV__sav_026796_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/886b3368c4b680bedf70/F0.jpg", "media/886b3368c4b680bedf70/F1.jpg", "media/886b3368c4b680bedf70/F2.jpg", "media/886b3368c4b680bedf70/F3.jpg", "media/886b3368c4b680bedf70/F4.jpg", "media/886b3368c4b680bedf70/F5.jpg", "media/886b3368c4b680bedf70/F6.jpg", "media/886b3368c4b680bedf70/F7.jpg", "media/886b3368c4b680bedf70/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "10f48f80ee457b4ad4a2ed68146a4ef8fa1445fc2b459caf8a9737d13f2a3be3"}
{"qa_id": "SAV__sav_026796_manual__camera_object_distance_change__02", "video_id": "SAV__sav_026796_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did object A (person in black wetsuit standing on a paddleboard) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/4d43a8332bd52ce25a9d/F0.jpg", "media/4d43a8332bd52ce25a9d/F1.jpg", "media/4d43a8332bd52ce25a9d/F2.jpg", "media/4d43a8332bd52ce25a9d/F3.jpg", "media/4d43a8332bd52ce25a9d/F4.jpg", "media/4d43a8332bd52ce25a9d/F5.jpg", "media/4d43a8332bd52ce25a9d/F6.jpg", "media/4d43a8332bd52ce25a9d/F7.jpg", "media/4d43a8332bd52ce25a9d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g5"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person in black wetsuit standing on a paddleboard) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dda686f70c98aa06907917225bfaa5ace359587099b06e4adbb72ee780593d5c"}
{"qa_id": "SAV__sav_026911_manual__camera_view_direction_change__00", "video_id": "SAV__sav_026911_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/8c255e2fb71b18baeb90/F0.jpg", "media/8c255e2fb71b18baeb90/F1.jpg", "media/8c255e2fb71b18baeb90/F2.jpg", "media/8c255e2fb71b18baeb90/F3.jpg", "media/8c255e2fb71b18baeb90/F4.jpg", "media/8c255e2fb71b18baeb90/F5.jpg", "media/8c255e2fb71b18baeb90/F6.jpg", "media/8c255e2fb71b18baeb90/F7.jpg", "media/8c255e2fb71b18baeb90/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b84a28e67f648a385add2ea863375298b93c1d91c4a839d56cf428371fa272d0"}
{"qa_id": "SAV__sav_026911_manual__object_world_motion_direction__01", "video_id": "SAV__sav_026911_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (small tan dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/70d9dfad24cc78f78d9e/F0.jpg", "media/70d9dfad24cc78f78d9e/F1.jpg", "media/70d9dfad24cc78f78d9e/F2.jpg", "media/70d9dfad24cc78f78d9e/F3.jpg", "media/70d9dfad24cc78f78d9e/F4.jpg", "media/70d9dfad24cc78f78d9e/F5.jpg", "media/70d9dfad24cc78f78d9e/F6.jpg", "media/70d9dfad24cc78f78d9e/F7.jpg", "media/70d9dfad24cc78f78d9e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (small tan dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "798dda9fd18734134857a40f05b914346b908b7379568d6c382432b4ecd53e17"}
{"qa_id": "SAV__sav_026911_manual__camera_object_distance_change__02", "video_id": "SAV__sav_026911_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (small tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/70d9dfad24cc78f78d9e/F0.jpg", "media/70d9dfad24cc78f78d9e/F1.jpg", "media/70d9dfad24cc78f78d9e/F2.jpg", "media/70d9dfad24cc78f78d9e/F3.jpg", "media/70d9dfad24cc78f78d9e/F4.jpg", "media/70d9dfad24cc78f78d9e/F5.jpg", "media/70d9dfad24cc78f78d9e/F6.jpg", "media/70d9dfad24cc78f78d9e/F7.jpg", "media/70d9dfad24cc78f78d9e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (small tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "451f3e2e0e4aea5ad2d94e44abc7b4dbe960b8d4242ddf15b03cd56925599439"}
{"qa_id": "SAV__sav_026977_manual__camera_object_distance_change__00", "video_id": "SAV__sav_026977_manual", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F7, did object A (black and tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/e1301ef64ce9a64886f1/F0.jpg", "media/e1301ef64ce9a64886f1/F1.jpg", "media/e1301ef64ce9a64886f1/F2.jpg", "media/e1301ef64ce9a64886f1/F3.jpg", "media/e1301ef64ce9a64886f1/F4.jpg", "media/e1301ef64ce9a64886f1/F5.jpg", "media/e1301ef64ce9a64886f1/F6.jpg", "media/e1301ef64ce9a64886f1/F7.jpg", "media/e1301ef64ce9a64886f1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did object A (black and tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2067a794f9d61bbdead502ed9c26b114d4ce17f7afd2de23f781d8591c4f8c7e"}
{"qa_id": "SAV__sav_026989_manual__camera_translation_direction__00", "video_id": "SAV__sav_026989_manual", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/a014fe87fba690d16395/F0.jpg", "media/a014fe87fba690d16395/F1.jpg", "media/a014fe87fba690d16395/F2.jpg", "media/a014fe87fba690d16395/F3.jpg", "media/a014fe87fba690d16395/F4.jpg", "media/a014fe87fba690d16395/F5.jpg", "media/a014fe87fba690d16395/F6.jpg", "media/a014fe87fba690d16395/F7.jpg", "media/a014fe87fba690d16395/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68f1ac42668e9b92932e4849c45d2560cccb93a6d0db8340ce64858d36df334d"}
{"qa_id": "SAV__sav_027268_manual__camera_translation_direction__00", "video_id": "SAV__sav_027268_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/aade100f3cae9f7eb020/F0.jpg", "media/aade100f3cae9f7eb020/F1.jpg", "media/aade100f3cae9f7eb020/F2.jpg", "media/aade100f3cae9f7eb020/F3.jpg", "media/aade100f3cae9f7eb020/F4.jpg", "media/aade100f3cae9f7eb020/F5.jpg", "media/aade100f3cae9f7eb020/F6.jpg", "media/aade100f3cae9f7eb020/F7.jpg", "media/aade100f3cae9f7eb020/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c09149ac5ab4e5a805c762ca98949b2b33dfd9c9fd7e33957ebbda0d9670766f"}
{"qa_id": "SAV__sav_027268_manual__object_world_motion_direction__01", "video_id": "SAV__sav_027268_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in green shirt riding a bicycle) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/862b5bcd160ff3545edb/F0.jpg", "media/862b5bcd160ff3545edb/F1.jpg", "media/862b5bcd160ff3545edb/F2.jpg", "media/862b5bcd160ff3545edb/F3.jpg", "media/862b5bcd160ff3545edb/F4.jpg", "media/862b5bcd160ff3545edb/F5.jpg", "media/862b5bcd160ff3545edb/F6.jpg", "media/862b5bcd160ff3545edb/F7.jpg", "media/862b5bcd160ff3545edb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in green shirt riding a bicycle) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f0bb97838cb7ecaaa44375f239565b994298e0a23d146af49895cb0852590722"}
{"qa_id": "SAV__sav_027604_manual__camera_translation_direction__00", "video_id": "SAV__sav_027604_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "up", "left", "backward", "forward", "down"], "frame_paths": ["media/c68793eac3d7ea173683/F0.jpg", "media/c68793eac3d7ea173683/F1.jpg", "media/c68793eac3d7ea173683/F2.jpg", "media/c68793eac3d7ea173683/F3.jpg", "media/c68793eac3d7ea173683/F4.jpg", "media/c68793eac3d7ea173683/F5.jpg", "media/c68793eac3d7ea173683/F6.jpg", "media/c68793eac3d7ea173683/F7.jpg", "media/c68793eac3d7ea173683/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f2f92621c393c5a31c1586f603625c22b64873021feb2a1414974d2045f619eb"}
{"qa_id": "SAV__sav_027651_manual__camera_translation_direction__00", "video_id": "SAV__sav_027651_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/daea8bb9b597c57b6b90/F0.jpg", "media/daea8bb9b597c57b6b90/F1.jpg", "media/daea8bb9b597c57b6b90/F2.jpg", "media/daea8bb9b597c57b6b90/F3.jpg", "media/daea8bb9b597c57b6b90/F4.jpg", "media/daea8bb9b597c57b6b90/F5.jpg", "media/daea8bb9b597c57b6b90/F6.jpg", "media/daea8bb9b597c57b6b90/F7.jpg", "media/daea8bb9b597c57b6b90/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b412ac774cc58251645df09f623d9defab604b235ddcb72f7490b32f131bc95c"}
{"qa_id": "SAV__sav_027651_manual__camera_view_direction_change__01", "video_id": "SAV__sav_027651_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/daea8bb9b597c57b6b90/F0.jpg", "media/daea8bb9b597c57b6b90/F1.jpg", "media/daea8bb9b597c57b6b90/F2.jpg", "media/daea8bb9b597c57b6b90/F3.jpg", "media/daea8bb9b597c57b6b90/F4.jpg", "media/daea8bb9b597c57b6b90/F5.jpg", "media/daea8bb9b597c57b6b90/F6.jpg", "media/daea8bb9b597c57b6b90/F7.jpg", "media/daea8bb9b597c57b6b90/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f98011f6f5e3bef7856bffbee3a8c2579537b239284b48df76fd5bbadf4bc3f"}
{"qa_id": "SAV__sav_027701_manual__camera_translation_direction__00", "video_id": "SAV__sav_027701_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/fc167427b06f1259fb5f/F0.jpg", "media/fc167427b06f1259fb5f/F1.jpg", "media/fc167427b06f1259fb5f/F2.jpg", "media/fc167427b06f1259fb5f/F3.jpg", "media/fc167427b06f1259fb5f/F4.jpg", "media/fc167427b06f1259fb5f/F5.jpg", "media/fc167427b06f1259fb5f/F6.jpg", "media/fc167427b06f1259fb5f/F7.jpg", "media/fc167427b06f1259fb5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78e426e1492f820c732e0a521f92efce55a392a87832866e919f54132d3fce03"}
{"qa_id": "SAV__sav_027701_manual__camera_view_direction_change__01", "video_id": "SAV__sav_027701_manual", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/fc167427b06f1259fb5f/F0.jpg", "media/fc167427b06f1259fb5f/F1.jpg", "media/fc167427b06f1259fb5f/F2.jpg", "media/fc167427b06f1259fb5f/F3.jpg", "media/fc167427b06f1259fb5f/F4.jpg", "media/fc167427b06f1259fb5f/F5.jpg", "media/fc167427b06f1259fb5f/F6.jpg", "media/fc167427b06f1259fb5f/F7.jpg", "media/fc167427b06f1259fb5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e94aa5aa390b996348d3ba18f896bc695065272669917f1a4e0987378a87ac1d"}
{"qa_id": "SAV__sav_027897_manual__camera_view_direction_change__00", "video_id": "SAV__sav_027897_manual", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/3021affb92d6a22b6acd/F0.jpg", "media/3021affb92d6a22b6acd/F1.jpg", "media/3021affb92d6a22b6acd/F2.jpg", "media/3021affb92d6a22b6acd/F3.jpg", "media/3021affb92d6a22b6acd/F4.jpg", "media/3021affb92d6a22b6acd/F5.jpg", "media/3021affb92d6a22b6acd/F6.jpg", "media/3021affb92d6a22b6acd/F7.jpg", "media/3021affb92d6a22b6acd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "SAV__sav_027897_manual__object_world_motion_direction__01", "video_id": "SAV__sav_027897_manual", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (yellow wheel loader) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/cabc3a721931f460f2db/F0.jpg", "media/cabc3a721931f460f2db/F1.jpg", "media/cabc3a721931f460f2db/F2.jpg", "media/cabc3a721931f460f2db/F3.jpg", "media/cabc3a721931f460f2db/F4.jpg", "media/cabc3a721931f460f2db/F5.jpg", "media/cabc3a721931f460f2db/F6.jpg", "media/cabc3a721931f460f2db/F7.jpg", "media/cabc3a721931f460f2db/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (yellow wheel loader) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "235329f9060023ef21f0b8706dd4c0b781eed1c67e93ec8beea8c77e7e30ae21"}
{"qa_id": "SAV__sav_027897_manual__object_travel_distance_comparison__02", "video_id": "SAV__sav_027897_manual", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (yellow wheel loader) or B (person in orange safety vest and white hard hat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/334f738af2b63fa5d921/F0.jpg", "media/334f738af2b63fa5d921/F1.jpg", "media/334f738af2b63fa5d921/F2.jpg", "media/334f738af2b63fa5d921/F3.jpg", "media/334f738af2b63fa5d921/F4.jpg", "media/334f738af2b63fa5d921/F5.jpg", "media/334f738af2b63fa5d921/F6.jpg", "media/334f738af2b63fa5d921/F7.jpg", "media/334f738af2b63fa5d921/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (yellow wheel loader) or B (person in orange safety vest and white hard hat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44e445352d4040f88ad0d6ce2d1b00dcb28fc0c0ee282f4cc16b2fc9a1a49f20"}
{"qa_id": "SAV__sav_028160_manual__object_world_motion_direction__00", "video_id": "SAV__sav_028160_manual", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (orange lion dance head) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/3e5122fc8217f487f034/F0.jpg", "media/3e5122fc8217f487f034/F1.jpg", "media/3e5122fc8217f487f034/F2.jpg", "media/3e5122fc8217f487f034/F3.jpg", "media/3e5122fc8217f487f034/F4.jpg", "media/3e5122fc8217f487f034/F5.jpg", "media/3e5122fc8217f487f034/F6.jpg", "media/3e5122fc8217f487f034/F7.jpg", "media/3e5122fc8217f487f034/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (orange lion dance head) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "17b2ee1a7e9ac7baef3e278cf5b23ee2beecac02c19da9a9e67cb6a9a1dd34eb"}
{"qa_id": "SAV__sav_028160_manual__object_pair_distance_change__01", "video_id": "SAV__sav_028160_manual", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (yellow banana) and object B (orange lion dance head) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ad16420faa08bfedf867/F0.jpg", "media/ad16420faa08bfedf867/F1.jpg", "media/ad16420faa08bfedf867/F2.jpg", "media/ad16420faa08bfedf867/F3.jpg", "media/ad16420faa08bfedf867/F4.jpg", "media/ad16420faa08bfedf867/F5.jpg", "media/ad16420faa08bfedf867/F6.jpg", "media/ad16420faa08bfedf867/F7.jpg", "media/ad16420faa08bfedf867/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (yellow banana) and object B (orange lion dance head) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6aca22fa7c05c62d51295ceee327408a786293e05f8ee6a2957542925d58c9c3"}
{"qa_id": "SAV__sav_028160_manual__object_travel_distance_comparison__02", "video_id": "SAV__sav_028160_manual", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (yellow banana) or B (orange lion dance head)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/ad16420faa08bfedf867/F0.jpg", "media/ad16420faa08bfedf867/F1.jpg", "media/ad16420faa08bfedf867/F2.jpg", "media/ad16420faa08bfedf867/F3.jpg", "media/ad16420faa08bfedf867/F4.jpg", "media/ad16420faa08bfedf867/F5.jpg", "media/ad16420faa08bfedf867/F6.jpg", "media/ad16420faa08bfedf867/F7.jpg", "media/ad16420faa08bfedf867/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (yellow banana) or B (orange lion dance head)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3caf770f51aac10e0e72bb63eed5f3179b5409d2d6c8111d225bbe8c2cbf4974"}
{"qa_id": "SAV__sav_028217_manual__camera_translation_direction__00", "video_id": "SAV__sav_028217_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/01f238b7f4f7f041136e/F0.jpg", "media/01f238b7f4f7f041136e/F1.jpg", "media/01f238b7f4f7f041136e/F2.jpg", "media/01f238b7f4f7f041136e/F3.jpg", "media/01f238b7f4f7f041136e/F4.jpg", "media/01f238b7f4f7f041136e/F5.jpg", "media/01f238b7f4f7f041136e/F6.jpg", "media/01f238b7f4f7f041136e/F7.jpg", "media/01f238b7f4f7f041136e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4bd1c1670851b2d6b80b3c195767d6f11c21a976148cd20ebeb51ee80d0fc7a5"}
{"qa_id": "SAV__sav_028284_manual__camera_translation_direction__00", "video_id": "SAV__sav_028284_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "right", "left", "forward"], "frame_paths": ["media/4f7be8ffda3e337103a5/F0.jpg", "media/4f7be8ffda3e337103a5/F1.jpg", "media/4f7be8ffda3e337103a5/F2.jpg", "media/4f7be8ffda3e337103a5/F3.jpg", "media/4f7be8ffda3e337103a5/F4.jpg", "media/4f7be8ffda3e337103a5/F5.jpg", "media/4f7be8ffda3e337103a5/F6.jpg", "media/4f7be8ffda3e337103a5/F7.jpg", "media/4f7be8ffda3e337103a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7a2ef02de7faaf4722ea0dd8e9b184d79af7dc66042c87ff37b4d7445ed9db40"}
{"qa_id": "SAV__sav_028284_manual__camera_view_direction_change__01", "video_id": "SAV__sav_028284_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/4f7be8ffda3e337103a5/F0.jpg", "media/4f7be8ffda3e337103a5/F1.jpg", "media/4f7be8ffda3e337103a5/F2.jpg", "media/4f7be8ffda3e337103a5/F3.jpg", "media/4f7be8ffda3e337103a5/F4.jpg", "media/4f7be8ffda3e337103a5/F5.jpg", "media/4f7be8ffda3e337103a5/F6.jpg", "media/4f7be8ffda3e337103a5/F7.jpg", "media/4f7be8ffda3e337103a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fca55b1611ae8dd2346bf922298894d229a7f1fe48f897ccdbc4dcbbbea48d9"}
{"qa_id": "SAV__sav_028490_manual__camera_translation_direction__00", "video_id": "SAV__sav_028490_manual", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/3b93f6f6564b5d7b198a/F0.jpg", "media/3b93f6f6564b5d7b198a/F1.jpg", "media/3b93f6f6564b5d7b198a/F2.jpg", "media/3b93f6f6564b5d7b198a/F3.jpg", "media/3b93f6f6564b5d7b198a/F4.jpg", "media/3b93f6f6564b5d7b198a/F5.jpg", "media/3b93f6f6564b5d7b198a/F6.jpg", "media/3b93f6f6564b5d7b198a/F7.jpg", "media/3b93f6f6564b5d7b198a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e6edb903a21201ec0802e9c5d7b43762951c682b85ba617d3457bf232c919b5"}
{"qa_id": "SAV__sav_028490_manual__object_world_motion_direction__01", "video_id": "SAV__sav_028490_manual", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (person in black coat carrying a large grey tote bag) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "right", "left", "forward"], "frame_paths": ["media/d7277882f64d62b722d8/F0.jpg", "media/d7277882f64d62b722d8/F1.jpg", "media/d7277882f64d62b722d8/F2.jpg", "media/d7277882f64d62b722d8/F3.jpg", "media/d7277882f64d62b722d8/F4.jpg", "media/d7277882f64d62b722d8/F5.jpg", "media/d7277882f64d62b722d8/F6.jpg", "media/d7277882f64d62b722d8/F7.jpg", "media/d7277882f64d62b722d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (person in black coat carrying a large grey tote bag) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c34a7d1c686a073582410fe6b174b1a70dff53f9be04765667ed1e2db2c25f3"}
{"qa_id": "SAV__sav_028490_manual__camera_object_distance_change__02", "video_id": "SAV__sav_028490_manual", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, did object A (person in black coat carrying a large grey tote bag) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d7277882f64d62b722d8/F0.jpg", "media/d7277882f64d62b722d8/F1.jpg", "media/d7277882f64d62b722d8/F2.jpg", "media/d7277882f64d62b722d8/F3.jpg", "media/d7277882f64d62b722d8/F4.jpg", "media/d7277882f64d62b722d8/F5.jpg", "media/d7277882f64d62b722d8/F6.jpg", "media/d7277882f64d62b722d8/F7.jpg", "media/d7277882f64d62b722d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in black coat carrying a large grey tote bag) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "018b4b30427d30af5235b4033157b14a29d80a66ae8def33e1c7c86cab8d8581"}
{"qa_id": "SAV__sav_028492_manual__camera_translation_direction__00", "video_id": "SAV__sav_028492_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/80456d2fd0981718f48f/F0.jpg", "media/80456d2fd0981718f48f/F1.jpg", "media/80456d2fd0981718f48f/F2.jpg", "media/80456d2fd0981718f48f/F3.jpg", "media/80456d2fd0981718f48f/F4.jpg", "media/80456d2fd0981718f48f/F5.jpg", "media/80456d2fd0981718f48f/F6.jpg", "media/80456d2fd0981718f48f/F7.jpg", "media/80456d2fd0981718f48f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "50618bba4266949699d6421a06cbd1d8b4ac196d01fb07ac036d0d33eaaf55b7"}
{"qa_id": "SAV__sav_028492_manual__camera_view_direction_change__01", "video_id": "SAV__sav_028492_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/80456d2fd0981718f48f/F0.jpg", "media/80456d2fd0981718f48f/F1.jpg", "media/80456d2fd0981718f48f/F2.jpg", "media/80456d2fd0981718f48f/F3.jpg", "media/80456d2fd0981718f48f/F4.jpg", "media/80456d2fd0981718f48f/F5.jpg", "media/80456d2fd0981718f48f/F6.jpg", "media/80456d2fd0981718f48f/F7.jpg", "media/80456d2fd0981718f48f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0ea2845daa9b9bbdef39a6303c9556d96c842c237c9934d2070b31ba0a0227eb"}
{"qa_id": "SAV__sav_028492_manual__object_world_motion_direction__02", "video_id": "SAV__sav_028492_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (person wearing a beige puffer jacket and pink pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "down", "up", "left"], "frame_paths": ["media/b83495bfd5248e1fa19f/F0.jpg", "media/b83495bfd5248e1fa19f/F1.jpg", "media/b83495bfd5248e1fa19f/F2.jpg", "media/b83495bfd5248e1fa19f/F3.jpg", "media/b83495bfd5248e1fa19f/F4.jpg", "media/b83495bfd5248e1fa19f/F5.jpg", "media/b83495bfd5248e1fa19f/F6.jpg", "media/b83495bfd5248e1fa19f/F7.jpg", "media/b83495bfd5248e1fa19f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (person wearing a beige puffer jacket and pink pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce46ea0929a9e5a82edb5c8a0f53016a3148183300d9b9b984e42db324a4d152"}
{"qa_id": "SAV__sav_028884_manual__object_world_motion_direction__00", "video_id": "SAV__sav_028884_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "forward", "down"], "frame_paths": ["media/789a894d17eabc864d57/F0.jpg", "media/789a894d17eabc864d57/F1.jpg", "media/789a894d17eabc864d57/F2.jpg", "media/789a894d17eabc864d57/F3.jpg", "media/789a894d17eabc864d57/F4.jpg", "media/789a894d17eabc864d57/F5.jpg", "media/789a894d17eabc864d57/F6.jpg", "media/789a894d17eabc864d57/F7.jpg", "media/789a894d17eabc864d57/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7a9da68b0e10048a2c6d7bb73a3e0ad9b5ef334bf0e6ca01eb55a8f5a816124f"}
{"qa_id": "SAV__sav_028884_manual__camera_object_distance_change__01", "video_id": "SAV__sav_028884_manual", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F3, did object A (person in orange shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/789a894d17eabc864d57/F0.jpg", "media/789a894d17eabc864d57/F1.jpg", "media/789a894d17eabc864d57/F2.jpg", "media/789a894d17eabc864d57/F3.jpg", "media/789a894d17eabc864d57/F4.jpg", "media/789a894d17eabc864d57/F5.jpg", "media/789a894d17eabc864d57/F6.jpg", "media/789a894d17eabc864d57/F7.jpg", "media/789a894d17eabc864d57/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in orange shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe61016572b4ac3017bc99be3fa204717492efb369b146feabb2d7abf5e7cb7c"}
{"qa_id": "SAV__sav_029089_manual__camera_translation_direction__00", "video_id": "SAV__sav_029089_manual", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "left", "right", "forward"], "frame_paths": ["media/941982abf0ae07330fc7/F0.jpg", "media/941982abf0ae07330fc7/F1.jpg", "media/941982abf0ae07330fc7/F2.jpg", "media/941982abf0ae07330fc7/F3.jpg", "media/941982abf0ae07330fc7/F4.jpg", "media/941982abf0ae07330fc7/F5.jpg", "media/941982abf0ae07330fc7/F6.jpg", "media/941982abf0ae07330fc7/F7.jpg", "media/941982abf0ae07330fc7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. left\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "337efe8eefc072665545c57c3f6df715c5195723919e06c866f09d650fb39156"}
{"qa_id": "SAV__sav_029089_manual__camera_view_direction_change__01", "video_id": "SAV__sav_029089_manual", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/941982abf0ae07330fc7/F0.jpg", "media/941982abf0ae07330fc7/F1.jpg", "media/941982abf0ae07330fc7/F2.jpg", "media/941982abf0ae07330fc7/F3.jpg", "media/941982abf0ae07330fc7/F4.jpg", "media/941982abf0ae07330fc7/F5.jpg", "media/941982abf0ae07330fc7/F6.jpg", "media/941982abf0ae07330fc7/F7.jpg", "media/941982abf0ae07330fc7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5bac693043594d1fe732e8d6be282a4d4e79d7e9c58acf4ef991c6952e1438e3"}
{"qa_id": "SAV__sav_029089_manual__object_world_motion_direction__02", "video_id": "SAV__sav_029089_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (black bag with white trim) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "backward", "forward", "down"], "frame_paths": ["media/612a2902db13e940befd/F0.jpg", "media/612a2902db13e940befd/F1.jpg", "media/612a2902db13e940befd/F2.jpg", "media/612a2902db13e940befd/F3.jpg", "media/612a2902db13e940befd/F4.jpg", "media/612a2902db13e940befd/F5.jpg", "media/612a2902db13e940befd/F6.jpg", "media/612a2902db13e940befd/F7.jpg", "media/612a2902db13e940befd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (black bag with white trim) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. backward\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fcbed9952ca41a26b7af2f977a97c6d0390d487bab2e654c0d754679f45c664a"}
{"qa_id": "SAV__sav_029089_manual__camera_object_distance_change__03", "video_id": "SAV__sav_029089_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object A (black bag with white trim) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/612a2902db13e940befd/F0.jpg", "media/612a2902db13e940befd/F1.jpg", "media/612a2902db13e940befd/F2.jpg", "media/612a2902db13e940befd/F3.jpg", "media/612a2902db13e940befd/F4.jpg", "media/612a2902db13e940befd/F5.jpg", "media/612a2902db13e940befd/F6.jpg", "media/612a2902db13e940befd/F7.jpg", "media/612a2902db13e940befd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (black bag with white trim) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2da8ce1a094224eaaf05535d15bd461dd6a7469e240118d90c1d7ae05c0778ac"}
{"qa_id": "SAV__sav_029089_manual__object_world_motion_direction__04", "video_id": "SAV__sav_029089_manual", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object B (toddler in striped shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "up", "down", "forward", "right"], "frame_paths": ["media/4b4809bd37d7afa66be8/F0.jpg", "media/4b4809bd37d7afa66be8/F1.jpg", "media/4b4809bd37d7afa66be8/F2.jpg", "media/4b4809bd37d7afa66be8/F3.jpg", "media/4b4809bd37d7afa66be8/F4.jpg", "media/4b4809bd37d7afa66be8/F5.jpg", "media/4b4809bd37d7afa66be8/F6.jpg", "media/4b4809bd37d7afa66be8/F7.jpg", "media/4b4809bd37d7afa66be8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object B (toddler in striped shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. up\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8827f1d427229d2385c410c60645993d0155e38409c84101109fb3e12a02b2b3"}
{"qa_id": "SAV__sav_029089_manual__camera_object_distance_change__05", "video_id": "SAV__sav_029089_manual", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object B (toddler in striped shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/4b4809bd37d7afa66be8/F0.jpg", "media/4b4809bd37d7afa66be8/F1.jpg", "media/4b4809bd37d7afa66be8/F2.jpg", "media/4b4809bd37d7afa66be8/F3.jpg", "media/4b4809bd37d7afa66be8/F4.jpg", "media/4b4809bd37d7afa66be8/F5.jpg", "media/4b4809bd37d7afa66be8/F6.jpg", "media/4b4809bd37d7afa66be8/F7.jpg", "media/4b4809bd37d7afa66be8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object B (toddler in striped shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f512684147ae74dfe8206274ec32c3b0f3a0c2d1ddcd05dc3419c76866ec367"}
{"qa_id": "SAV__sav_029393_manual__camera_translation_direction__00", "video_id": "SAV__sav_029393_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/1abd77a62026e6ad85e6/F0.jpg", "media/1abd77a62026e6ad85e6/F1.jpg", "media/1abd77a62026e6ad85e6/F2.jpg", "media/1abd77a62026e6ad85e6/F3.jpg", "media/1abd77a62026e6ad85e6/F4.jpg", "media/1abd77a62026e6ad85e6/F5.jpg", "media/1abd77a62026e6ad85e6/F6.jpg", "media/1abd77a62026e6ad85e6/F7.jpg", "media/1abd77a62026e6ad85e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6dc67d45f1fb12b306f3d07d5424cdfa935f44794adb849f2c65064798657790"}
{"qa_id": "SAV__sav_029393_manual__camera_view_direction_change__01", "video_id": "SAV__sav_029393_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/1abd77a62026e6ad85e6/F0.jpg", "media/1abd77a62026e6ad85e6/F1.jpg", "media/1abd77a62026e6ad85e6/F2.jpg", "media/1abd77a62026e6ad85e6/F3.jpg", "media/1abd77a62026e6ad85e6/F4.jpg", "media/1abd77a62026e6ad85e6/F5.jpg", "media/1abd77a62026e6ad85e6/F6.jpg", "media/1abd77a62026e6ad85e6/F7.jpg", "media/1abd77a62026e6ad85e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e5a587eeadc737b189dad5b9cd545c84273184484229581ea01a145a6f9ce13"}
{"qa_id": "SAV__sav_029393_manual__object_world_motion_direction__02", "video_id": "SAV__sav_029393_manual", "task": "object_world_motion_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "down", "left", "right", "forward"], "frame_paths": ["media/4b69eefc9c12bef1a85f/F0.jpg", "media/4b69eefc9c12bef1a85f/F1.jpg", "media/4b69eefc9c12bef1a85f/F2.jpg", "media/4b69eefc9c12bef1a85f/F3.jpg", "media/4b69eefc9c12bef1a85f/F4.jpg", "media/4b69eefc9c12bef1a85f/F5.jpg", "media/4b69eefc9c12bef1a85f/F6.jpg", "media/4b69eefc9c12bef1a85f/F7.jpg", "media/4b69eefc9c12bef1a85f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. down\nD. left\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f501501e1af814a3926e7cdcd5189fb2483d8ef27a804a41525395e9f5d8a0b"}
{"qa_id": "SAV__sav_029396_manual__camera_translation_direction__00", "video_id": "SAV__sav_029396_manual", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/abc5a54c821afeb140c3/F0.jpg", "media/abc5a54c821afeb140c3/F1.jpg", "media/abc5a54c821afeb140c3/F2.jpg", "media/abc5a54c821afeb140c3/F3.jpg", "media/abc5a54c821afeb140c3/F4.jpg", "media/abc5a54c821afeb140c3/F5.jpg", "media/abc5a54c821afeb140c3/F6.jpg", "media/abc5a54c821afeb140c3/F7.jpg", "media/abc5a54c821afeb140c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68f1ac42668e9b92932e4849c45d2560cccb93a6d0db8340ce64858d36df334d"}
{"qa_id": "SAV__sav_029396_manual__camera_view_direction_change__01", "video_id": "SAV__sav_029396_manual", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/abc5a54c821afeb140c3/F0.jpg", "media/abc5a54c821afeb140c3/F1.jpg", "media/abc5a54c821afeb140c3/F2.jpg", "media/abc5a54c821afeb140c3/F3.jpg", "media/abc5a54c821afeb140c3/F4.jpg", "media/abc5a54c821afeb140c3/F5.jpg", "media/abc5a54c821afeb140c3/F6.jpg", "media/abc5a54c821afeb140c3/F7.jpg", "media/abc5a54c821afeb140c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e35be618d373050de58880ab60b6f867433ccbfc620057e6b5f7be2d977f2eb3"}
{"qa_id": "SAV__sav_029473_manual__object_world_motion_direction__00", "video_id": "SAV__sav_029473_manual", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in dark clothing walking on the path) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/39df414432470fd736d2/F0.jpg", "media/39df414432470fd736d2/F1.jpg", "media/39df414432470fd736d2/F2.jpg", "media/39df414432470fd736d2/F3.jpg", "media/39df414432470fd736d2/F4.jpg", "media/39df414432470fd736d2/F5.jpg", "media/39df414432470fd736d2/F6.jpg", "media/39df414432470fd736d2/F7.jpg", "media/39df414432470fd736d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in dark clothing walking on the path) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "659316e194720fcb5767b31eac2e6b6bb487223b59433c21cf7d39db5371d111"}
{"qa_id": "SAV__sav_029594_manual__camera_translation_direction__00", "video_id": "SAV__sav_029594_manual", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/ec410f90c49e3ea743cb/F0.jpg", "media/ec410f90c49e3ea743cb/F1.jpg", "media/ec410f90c49e3ea743cb/F2.jpg", "media/ec410f90c49e3ea743cb/F3.jpg", "media/ec410f90c49e3ea743cb/F4.jpg", "media/ec410f90c49e3ea743cb/F5.jpg", "media/ec410f90c49e3ea743cb/F6.jpg", "media/ec410f90c49e3ea743cb/F7.jpg", "media/ec410f90c49e3ea743cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68f1ac42668e9b92932e4849c45d2560cccb93a6d0db8340ce64858d36df334d"}
{"qa_id": "SAV__sav_029594_manual__camera_view_direction_change__01", "video_id": "SAV__sav_029594_manual", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/ec410f90c49e3ea743cb/F0.jpg", "media/ec410f90c49e3ea743cb/F1.jpg", "media/ec410f90c49e3ea743cb/F2.jpg", "media/ec410f90c49e3ea743cb/F3.jpg", "media/ec410f90c49e3ea743cb/F4.jpg", "media/ec410f90c49e3ea743cb/F5.jpg", "media/ec410f90c49e3ea743cb/F6.jpg", "media/ec410f90c49e3ea743cb/F7.jpg", "media/ec410f90c49e3ea743cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe49dec5b326f5c289d8be859ecbaac9b78bbdb988deb8691948135fcd374071"}
{"qa_id": "SAV__sav_029594_manual__camera_object_distance_change__02", "video_id": "SAV__sav_029594_manual", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, did object A (baby in pink outfit sitting in a stroller) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c3e2d2b5941a1fdb6ec8/F0.jpg", "media/c3e2d2b5941a1fdb6ec8/F1.jpg", "media/c3e2d2b5941a1fdb6ec8/F2.jpg", "media/c3e2d2b5941a1fdb6ec8/F3.jpg", "media/c3e2d2b5941a1fdb6ec8/F4.jpg", "media/c3e2d2b5941a1fdb6ec8/F5.jpg", "media/c3e2d2b5941a1fdb6ec8/F6.jpg", "media/c3e2d2b5941a1fdb6ec8/F7.jpg", "media/c3e2d2b5941a1fdb6ec8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (baby in pink outfit sitting in a stroller) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e985ce752b76d974a4a7e791438a3fa59ff497a5264d3166ae2df6833ed22ab7"}
{"qa_id": "VOST__1005_cut_potato__camera_translation_direction__00", "video_id": "VOST__1005_cut_potato", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "down", "forward", "left", "up"], "frame_paths": ["media/295ce5644ae9c50be6e5/F0.jpg", "media/295ce5644ae9c50be6e5/F1.jpg", "media/295ce5644ae9c50be6e5/F2.jpg", "media/295ce5644ae9c50be6e5/F3.jpg", "media/295ce5644ae9c50be6e5/F4.jpg", "media/295ce5644ae9c50be6e5/F5.jpg", "media/295ce5644ae9c50be6e5/F6.jpg", "media/295ce5644ae9c50be6e5/F7.jpg", "media/295ce5644ae9c50be6e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. down\nD. forward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a60b7d2e4ddd5d0d97f125ab59a0efe28c8034b98c3db73a0a92e1254620afbb"}
{"qa_id": "VOST__1005_cut_potato__camera_view_direction_change__01", "video_id": "VOST__1005_cut_potato", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/295ce5644ae9c50be6e5/F0.jpg", "media/295ce5644ae9c50be6e5/F1.jpg", "media/295ce5644ae9c50be6e5/F2.jpg", "media/295ce5644ae9c50be6e5/F3.jpg", "media/295ce5644ae9c50be6e5/F4.jpg", "media/295ce5644ae9c50be6e5/F5.jpg", "media/295ce5644ae9c50be6e5/F6.jpg", "media/295ce5644ae9c50be6e5/F7.jpg", "media/295ce5644ae9c50be6e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6e7548e01f85a3b9bc6ccaa1c81ba06affcc5c5c3f994721885637ea7b3a55da"}
{"qa_id": "VOST__1006_cut_onion__camera_translation_direction__00", "video_id": "VOST__1006_cut_onion", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/efdf5b382368061d2a93/F0.jpg", "media/efdf5b382368061d2a93/F1.jpg", "media/efdf5b382368061d2a93/F2.jpg", "media/efdf5b382368061d2a93/F3.jpg", "media/efdf5b382368061d2a93/F4.jpg", "media/efdf5b382368061d2a93/F5.jpg", "media/efdf5b382368061d2a93/F6.jpg", "media/efdf5b382368061d2a93/F7.jpg", "media/efdf5b382368061d2a93/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38346ba1f9a34aac5da7cdc2ca914ec44b54cdc29a108ad0ae7dec4ba4048b29"}
{"qa_id": "VOST__1008_peel_cabbage__camera_translation_direction__00", "video_id": "VOST__1008_peel_cabbage", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "forward", "down", "right"], "frame_paths": ["media/96b6ba1ab166e6bafbbd/F0.jpg", "media/96b6ba1ab166e6bafbbd/F1.jpg", "media/96b6ba1ab166e6bafbbd/F2.jpg", "media/96b6ba1ab166e6bafbbd/F3.jpg", "media/96b6ba1ab166e6bafbbd/F4.jpg", "media/96b6ba1ab166e6bafbbd/F5.jpg", "media/96b6ba1ab166e6bafbbd/F6.jpg", "media/96b6ba1ab166e6bafbbd/F7.jpg", "media/96b6ba1ab166e6bafbbd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. forward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9bb0f6fa7002be0d3404ca69955cdca2783fb4a3b659062abff1570c2e754827"}
{"qa_id": "VOST__10343_smooth_wood__camera_object_distance_change__00", "video_id": "VOST__10343_smooth_wood", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, did object B (person's left arm wearing a blue glove and a red wristband) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/ad542fc248715d41d160/F0.jpg", "media/ad542fc248715d41d160/F1.jpg", "media/ad542fc248715d41d160/F2.jpg", "media/ad542fc248715d41d160/F3.jpg", "media/ad542fc248715d41d160/F4.jpg", "media/ad542fc248715d41d160/F5.jpg", "media/ad542fc248715d41d160/F6.jpg", "media/ad542fc248715d41d160/F7.jpg", "media/ad542fc248715d41d160/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, did object B (person's left arm wearing a blue glove and a red wristband) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7a56765dcf49482bd05eddd9b66dd7ffceffaf3e2b3b8c3924324112a73d9e1e"}
{"qa_id": "VOST__10364_grate_carrot__camera_translation_direction__00", "video_id": "VOST__10364_grate_carrot", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/cadfd157774bebde3e36/F0.jpg", "media/cadfd157774bebde3e36/F1.jpg", "media/cadfd157774bebde3e36/F2.jpg", "media/cadfd157774bebde3e36/F3.jpg", "media/cadfd157774bebde3e36/F4.jpg", "media/cadfd157774bebde3e36/F5.jpg", "media/cadfd157774bebde3e36/F6.jpg", "media/cadfd157774bebde3e36/F7.jpg", "media/cadfd157774bebde3e36/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4dd5750d2b1498617a0e867d02ecc0097ca868ab28f49107d979e34fc39f8433"}
{"qa_id": "VOST__10364_grate_carrot__object_world_motion_direction__01", "video_id": "VOST__10364_grate_carrot", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (left arm of a person wearing a white t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "up", "forward", "left"], "frame_paths": ["media/32f045ce49927eb3aaa2/F0.jpg", "media/32f045ce49927eb3aaa2/F1.jpg", "media/32f045ce49927eb3aaa2/F2.jpg", "media/32f045ce49927eb3aaa2/F3.jpg", "media/32f045ce49927eb3aaa2/F4.jpg", "media/32f045ce49927eb3aaa2/F5.jpg", "media/32f045ce49927eb3aaa2/F6.jpg", "media/32f045ce49927eb3aaa2/F7.jpg", "media/32f045ce49927eb3aaa2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (left arm of a person wearing a white t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "feadeaa4ba878f289b1524c0662d8d5abf6842a7ee465e7e283fde42c0bdfac6"}
{"qa_id": "VOST__10364_grate_carrot__object_pair_distance_change__02", "video_id": "VOST__10364_grate_carrot", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (left arm of a person wearing a white t-shirt) and object B (right arm and torso of a person wearing a white t-shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ab4bf14cb2a064d46876/F0.jpg", "media/ab4bf14cb2a064d46876/F1.jpg", "media/ab4bf14cb2a064d46876/F2.jpg", "media/ab4bf14cb2a064d46876/F3.jpg", "media/ab4bf14cb2a064d46876/F4.jpg", "media/ab4bf14cb2a064d46876/F5.jpg", "media/ab4bf14cb2a064d46876/F6.jpg", "media/ab4bf14cb2a064d46876/F7.jpg", "media/ab4bf14cb2a064d46876/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (left arm of a person wearing a white t-shirt) and object B (right arm and torso of a person wearing a white t-shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9589276bb2dd6db56a61d3f8c0d2a6e8a49a0a375268b63b2be0e2f32a7f38cd"}
{"qa_id": "VOST__10364_grate_carrot__object_travel_distance_comparison__03", "video_id": "VOST__10364_grate_carrot", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm of a person wearing a white t-shirt) or B (right arm and torso of a person wearing a white t-shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/ab4bf14cb2a064d46876/F0.jpg", "media/ab4bf14cb2a064d46876/F1.jpg", "media/ab4bf14cb2a064d46876/F2.jpg", "media/ab4bf14cb2a064d46876/F3.jpg", "media/ab4bf14cb2a064d46876/F4.jpg", "media/ab4bf14cb2a064d46876/F5.jpg", "media/ab4bf14cb2a064d46876/F6.jpg", "media/ab4bf14cb2a064d46876/F7.jpg", "media/ab4bf14cb2a064d46876/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm of a person wearing a white t-shirt) or B (right arm and torso of a person wearing a white t-shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa28c586f98fc9f682f1d33a839eeab544323b9d2ef4bc9aa831d7ff5c656946"}
{"qa_id": "VOST__10379_sand_wood__camera_translation_direction__00", "video_id": "VOST__10379_sand_wood", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "backward", "right", "forward"], "frame_paths": ["media/755a7821940896062f2e/F0.jpg", "media/755a7821940896062f2e/F1.jpg", "media/755a7821940896062f2e/F2.jpg", "media/755a7821940896062f2e/F3.jpg", "media/755a7821940896062f2e/F4.jpg", "media/755a7821940896062f2e/F5.jpg", "media/755a7821940896062f2e/F6.jpg", "media/755a7821940896062f2e/F7.jpg", "media/755a7821940896062f2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. backward\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b3cf26dbf57a9f3a92889b1eb3b1a1e5dcdca2b3cdd88d8e9041207cdf4a9a12"}
{"qa_id": "VOST__10379_sand_wood__camera_view_direction_change__01", "video_id": "VOST__10379_sand_wood", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/755a7821940896062f2e/F0.jpg", "media/755a7821940896062f2e/F1.jpg", "media/755a7821940896062f2e/F2.jpg", "media/755a7821940896062f2e/F3.jpg", "media/755a7821940896062f2e/F4.jpg", "media/755a7821940896062f2e/F5.jpg", "media/755a7821940896062f2e/F6.jpg", "media/755a7821940896062f2e/F7.jpg", "media/755a7821940896062f2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82fb06dddece22d028c76a1de500fe7f0467ea6c1091f07daafc7c36a5581ad7"}
{"qa_id": "VOST__10574_weld_iron__camera_object_distance_change__01", "video_id": "VOST__10574_weld_iron", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object B (person wearing blue jeans and brown sandals, holding a hammer) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/8de1d5a22015ef9b65b1/F0.jpg", "media/8de1d5a22015ef9b65b1/F1.jpg", "media/8de1d5a22015ef9b65b1/F2.jpg", "media/8de1d5a22015ef9b65b1/F3.jpg", "media/8de1d5a22015ef9b65b1/F4.jpg", "media/8de1d5a22015ef9b65b1/F5.jpg", "media/8de1d5a22015ef9b65b1/F6.jpg", "media/8de1d5a22015ef9b65b1/F7.jpg", "media/8de1d5a22015ef9b65b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (person wearing blue jeans and brown sandals, holding a hammer) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "540b6cabf9349f565fe9584e46be2c2dc53ed46ad933d9e9401c45d1d1f982b7"}
{"qa_id": "VOST__10624_separate_dough__object_world_motion_direction__00", "video_id": "VOST__10624_separate_dough", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (right arm wearing a blue nitrile glove) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "left", "forward", "up", "down"], "frame_paths": ["media/8db1384c9301b40ac5cf/F0.jpg", "media/8db1384c9301b40ac5cf/F1.jpg", "media/8db1384c9301b40ac5cf/F2.jpg", "media/8db1384c9301b40ac5cf/F3.jpg", "media/8db1384c9301b40ac5cf/F4.jpg", "media/8db1384c9301b40ac5cf/F5.jpg", "media/8db1384c9301b40ac5cf/F6.jpg", "media/8db1384c9301b40ac5cf/F7.jpg", "media/8db1384c9301b40ac5cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (right arm wearing a blue nitrile glove) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. left\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f4321c725b719768fd8a7e4fc127464ce46cb315b35c07b44e16e329ea7f650b"}
{"qa_id": "VOST__10624_separate_dough__object_pair_distance_change__01", "video_id": "VOST__10624_separate_dough", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, did the 3D distance between object A (right arm wearing a blue nitrile glove) and object B (left arm wearing a blue nitrile glove) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/0a80693609816043f9f8/F0.jpg", "media/0a80693609816043f9f8/F1.jpg", "media/0a80693609816043f9f8/F2.jpg", "media/0a80693609816043f9f8/F3.jpg", "media/0a80693609816043f9f8/F4.jpg", "media/0a80693609816043f9f8/F5.jpg", "media/0a80693609816043f9f8/F6.jpg", "media/0a80693609816043f9f8/F7.jpg", "media/0a80693609816043f9f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (right arm wearing a blue nitrile glove) and object B (left arm wearing a blue nitrile glove) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3b9928ddfe83707394b925fe91c996fe0dd9bd7b76a0c62882f864b77ae55f6d"}
{"qa_id": "VOST__10698_cut_paper__camera_translation_direction__00", "video_id": "VOST__10698_cut_paper", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "left", "right", "backward"], "frame_paths": ["media/69c7b5f18a666f2e974c/F0.jpg", "media/69c7b5f18a666f2e974c/F1.jpg", "media/69c7b5f18a666f2e974c/F2.jpg", "media/69c7b5f18a666f2e974c/F3.jpg", "media/69c7b5f18a666f2e974c/F4.jpg", "media/69c7b5f18a666f2e974c/F5.jpg", "media/69c7b5f18a666f2e974c/F6.jpg", "media/69c7b5f18a666f2e974c/F7.jpg", "media/69c7b5f18a666f2e974c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c25748e8b106f6459da0e471901015cab7f390745e6d27113dcf8aab144c19e8"}
{"qa_id": "VOST__10698_cut_paper__object_pair_distance_change__01", "video_id": "VOST__10698_cut_paper", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did the 3D distance between object A (right arm and hand holding scissors) and object B (left arm and hand holding paper) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/8c684275fe4b9488ead4/F0.jpg", "media/8c684275fe4b9488ead4/F1.jpg", "media/8c684275fe4b9488ead4/F2.jpg", "media/8c684275fe4b9488ead4/F3.jpg", "media/8c684275fe4b9488ead4/F4.jpg", "media/8c684275fe4b9488ead4/F5.jpg", "media/8c684275fe4b9488ead4/F6.jpg", "media/8c684275fe4b9488ead4/F7.jpg", "media/8c684275fe4b9488ead4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (right arm and hand holding scissors) and object B (left arm and hand holding paper) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e58b9032d9412d69a3503cb8d3f5b99b0e50b3bba0333713d6e208ab09b080a1"}
{"qa_id": "VOST__10698_cut_paper__object_travel_distance_comparison__02", "video_id": "VOST__10698_cut_paper", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right arm and hand holding scissors) or B (left arm and hand holding paper)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/8c684275fe4b9488ead4/F0.jpg", "media/8c684275fe4b9488ead4/F1.jpg", "media/8c684275fe4b9488ead4/F2.jpg", "media/8c684275fe4b9488ead4/F3.jpg", "media/8c684275fe4b9488ead4/F4.jpg", "media/8c684275fe4b9488ead4/F5.jpg", "media/8c684275fe4b9488ead4/F6.jpg", "media/8c684275fe4b9488ead4/F7.jpg", "media/8c684275fe4b9488ead4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right arm and hand holding scissors) or B (left arm and hand holding paper)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "77544f9acf86c41f9de4cfb80a9d48d5d644d05775d304c7eb4c6c247366303c"}
{"qa_id": "VOST__10701_cut_cake__camera_translation_direction__00", "video_id": "VOST__10701_cut_cake", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "down", "up", "forward"], "frame_paths": ["media/6dac3a90a2905658fe72/F0.jpg", "media/6dac3a90a2905658fe72/F1.jpg", "media/6dac3a90a2905658fe72/F2.jpg", "media/6dac3a90a2905658fe72/F3.jpg", "media/6dac3a90a2905658fe72/F4.jpg", "media/6dac3a90a2905658fe72/F5.jpg", "media/6dac3a90a2905658fe72/F6.jpg", "media/6dac3a90a2905658fe72/F7.jpg", "media/6dac3a90a2905658fe72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7d9c9d88cf635870383fed1068493142db3ca0043b81aabd7552df4b1734b5a8"}
{"qa_id": "VOST__1082_cut_tomato__camera_translation_direction__00", "video_id": "VOST__1082_cut_tomato", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "left", "backward", "up"], "frame_paths": ["media/818f4464ae56ae419aa6/F0.jpg", "media/818f4464ae56ae419aa6/F1.jpg", "media/818f4464ae56ae419aa6/F2.jpg", "media/818f4464ae56ae419aa6/F3.jpg", "media/818f4464ae56ae419aa6/F4.jpg", "media/818f4464ae56ae419aa6/F5.jpg", "media/818f4464ae56ae419aa6/F6.jpg", "media/818f4464ae56ae419aa6/F7.jpg", "media/818f4464ae56ae419aa6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. left\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4b8bad0baa3c2681eb9d748c8701a472316985a4683133d6649c1d323fbf3a2"}
{"qa_id": "VOST__1186_cut_chilli__object_world_motion_direction__00", "video_id": "VOST__1186_cut_chilli", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (left hand and forearm of a person, holding food on a cutting board) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "forward", "backward", "left"], "frame_paths": ["media/1d58659bc236b31f4f02/F0.jpg", "media/1d58659bc236b31f4f02/F1.jpg", "media/1d58659bc236b31f4f02/F2.jpg", "media/1d58659bc236b31f4f02/F3.jpg", "media/1d58659bc236b31f4f02/F4.jpg", "media/1d58659bc236b31f4f02/F5.jpg", "media/1d58659bc236b31f4f02/F6.jpg", "media/1d58659bc236b31f4f02/F7.jpg", "media/1d58659bc236b31f4f02/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (left hand and forearm of a person, holding food on a cutting board) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3624698e942d89415c7d78dd5c1050d97bf2aaf5313b4fb79cff4ec40352fc69"}
{"qa_id": "VOST__1186_cut_chilli__object_pair_distance_change__01", "video_id": "VOST__1186_cut_chilli", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did the 3D distance between object A (right hand and forearm of a person wearing a dark wristwatch, holding a knife) and object B (left hand and forearm of a person, holding food on a cutting board) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/a455de2da4d8e788500e/F0.jpg", "media/a455de2da4d8e788500e/F1.jpg", "media/a455de2da4d8e788500e/F2.jpg", "media/a455de2da4d8e788500e/F3.jpg", "media/a455de2da4d8e788500e/F4.jpg", "media/a455de2da4d8e788500e/F5.jpg", "media/a455de2da4d8e788500e/F6.jpg", "media/a455de2da4d8e788500e/F7.jpg", "media/a455de2da4d8e788500e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (right hand and forearm of a person wearing a dark wristwatch, holding a knife) and object B (left hand and forearm of a person, holding food on a cutting board) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "df400ef82a8fc2be61686063d47c4798e28d5ab4b2c78337100a64aa3eb835c0"}
{"qa_id": "VOST__1186_cut_chilli__object_travel_distance_comparison__02", "video_id": "VOST__1186_cut_chilli", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand and forearm of a person wearing a dark wristwatch, holding a knife) or B (left hand and forearm of a person, holding food on a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/a455de2da4d8e788500e/F0.jpg", "media/a455de2da4d8e788500e/F1.jpg", "media/a455de2da4d8e788500e/F2.jpg", "media/a455de2da4d8e788500e/F3.jpg", "media/a455de2da4d8e788500e/F4.jpg", "media/a455de2da4d8e788500e/F5.jpg", "media/a455de2da4d8e788500e/F6.jpg", "media/a455de2da4d8e788500e/F7.jpg", "media/a455de2da4d8e788500e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand and forearm of a person wearing a dark wristwatch, holding a knife) or B (left hand and forearm of a person, holding food on a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dfc81b330c1ff615fc663f5856c6221fc588905383429094f966a5bf2bb0dd94"}
{"qa_id": "VOST__1200_cut_onion__camera_translation_direction__00", "video_id": "VOST__1200_cut_onion", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "down", "up", "left", "forward"], "frame_paths": ["media/22d537d48fdb21a62e48/F0.jpg", "media/22d537d48fdb21a62e48/F1.jpg", "media/22d537d48fdb21a62e48/F2.jpg", "media/22d537d48fdb21a62e48/F3.jpg", "media/22d537d48fdb21a62e48/F4.jpg", "media/22d537d48fdb21a62e48/F5.jpg", "media/22d537d48fdb21a62e48/F6.jpg", "media/22d537d48fdb21a62e48/F7.jpg", "media/22d537d48fdb21a62e48/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. down\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "11ceb24489fc724e8113433d71477c9b80020ec67d1285b63bfee8f400a5da1a"}
{"qa_id": "VOST__1292_cut_cake__camera_translation_direction__00", "video_id": "VOST__1292_cut_cake", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "down", "forward", "right", "backward"], "frame_paths": ["media/70dffe36a8c6d42c3d63/F0.jpg", "media/70dffe36a8c6d42c3d63/F1.jpg", "media/70dffe36a8c6d42c3d63/F2.jpg", "media/70dffe36a8c6d42c3d63/F3.jpg", "media/70dffe36a8c6d42c3d63/F4.jpg", "media/70dffe36a8c6d42c3d63/F5.jpg", "media/70dffe36a8c6d42c3d63/F6.jpg", "media/70dffe36a8c6d42c3d63/F7.jpg", "media/70dffe36a8c6d42c3d63/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. down\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5b4741ea79642bca687f3039809228ff982efabf9474e43f5dfc1fc4d626b165"}
{"qa_id": "VOST__1293_cut_mozzarella__camera_translation_direction__00", "video_id": "VOST__1293_cut_mozzarella", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "down", "backward", "left", "forward", "up"], "frame_paths": ["media/4712c4f3ba0e24d25a73/F0.jpg", "media/4712c4f3ba0e24d25a73/F1.jpg", "media/4712c4f3ba0e24d25a73/F2.jpg", "media/4712c4f3ba0e24d25a73/F3.jpg", "media/4712c4f3ba0e24d25a73/F4.jpg", "media/4712c4f3ba0e24d25a73/F5.jpg", "media/4712c4f3ba0e24d25a73/F6.jpg", "media/4712c4f3ba0e24d25a73/F7.jpg", "media/4712c4f3ba0e24d25a73/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. backward\nD. left\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7095aa947fae02681c2d459040592d995a4f936e13d8cd9fcd1d2690a67eed20"}
{"qa_id": "VOST__1293_cut_mozzarella__object_pair_distance_change__01", "video_id": "VOST__1293_cut_mozzarella", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did the 3D distance between object A (right hand holding a knife) and object B (left hand holding a white object) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/e5d2388e910091bb5181/F0.jpg", "media/e5d2388e910091bb5181/F1.jpg", "media/e5d2388e910091bb5181/F2.jpg", "media/e5d2388e910091bb5181/F3.jpg", "media/e5d2388e910091bb5181/F4.jpg", "media/e5d2388e910091bb5181/F5.jpg", "media/e5d2388e910091bb5181/F6.jpg", "media/e5d2388e910091bb5181/F7.jpg", "media/e5d2388e910091bb5181/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (right hand holding a knife) and object B (left hand holding a white object) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "86f695deedf3a0dada49976733be58bd0ab2bd7c780c2c5abf9a9830a0d7e4e1"}
{"qa_id": "VOST__1389_peel_banana__camera_translation_direction__00", "video_id": "VOST__1389_peel_banana", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "backward", "down", "up"], "frame_paths": ["media/10b334e2f9bbeff6cc02/F0.jpg", "media/10b334e2f9bbeff6cc02/F1.jpg", "media/10b334e2f9bbeff6cc02/F2.jpg", "media/10b334e2f9bbeff6cc02/F3.jpg", "media/10b334e2f9bbeff6cc02/F4.jpg", "media/10b334e2f9bbeff6cc02/F5.jpg", "media/10b334e2f9bbeff6cc02/F6.jpg", "media/10b334e2f9bbeff6cc02/F7.jpg", "media/10b334e2f9bbeff6cc02/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab54feeb6754eaaa6992163e11e268b4d7f70002c2bacb7460ebada5763f24b9"}
{"qa_id": "VOST__1398_cut_ginger__camera_translation_direction__00", "video_id": "VOST__1398_cut_ginger", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/ddc6a419f3c71a986f41/F0.jpg", "media/ddc6a419f3c71a986f41/F1.jpg", "media/ddc6a419f3c71a986f41/F2.jpg", "media/ddc6a419f3c71a986f41/F3.jpg", "media/ddc6a419f3c71a986f41/F4.jpg", "media/ddc6a419f3c71a986f41/F5.jpg", "media/ddc6a419f3c71a986f41/F6.jpg", "media/ddc6a419f3c71a986f41/F7.jpg", "media/ddc6a419f3c71a986f41/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "01f9b1002e8fd901003d5b8e9462824cb5b6da382b70db6af951c0edbb330318"}
{"qa_id": "VOST__1445_cut_carrot__object_travel_distance_comparison__00", "video_id": "VOST__1445_cut_carrot", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding a carrot)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/334dd71b0ba7901c8586/F0.jpg", "media/334dd71b0ba7901c8586/F1.jpg", "media/334dd71b0ba7901c8586/F2.jpg", "media/334dd71b0ba7901c8586/F3.jpg", "media/334dd71b0ba7901c8586/F4.jpg", "media/334dd71b0ba7901c8586/F5.jpg", "media/334dd71b0ba7901c8586/F6.jpg", "media/334dd71b0ba7901c8586/F7.jpg", "media/334dd71b0ba7901c8586/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding a carrot)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d1a977f85cf165741919671b1f70f98cfc41e3b0bcbcf6ab7e892b947104366"}
{"qa_id": "VOST__1454_cut_carrot__object_travel_distance_comparison__00", "video_id": "VOST__1454_cut_carrot", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding a carrot)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/6a49f8f1364366e4aff0/F0.jpg", "media/6a49f8f1364366e4aff0/F1.jpg", "media/6a49f8f1364366e4aff0/F2.jpg", "media/6a49f8f1364366e4aff0/F3.jpg", "media/6a49f8f1364366e4aff0/F4.jpg", "media/6a49f8f1364366e4aff0/F5.jpg", "media/6a49f8f1364366e4aff0/F6.jpg", "media/6a49f8f1364366e4aff0/F7.jpg", "media/6a49f8f1364366e4aff0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding a carrot)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "de07e8433db9ddde3e3d8bd5e44d79274c477f7390112eba286cc6716ba8554e"}
{"qa_id": "VOST__1456_cut_carrot__camera_translation_direction__00", "video_id": "VOST__1456_cut_carrot", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "right", "forward"], "frame_paths": ["media/e3c46a5a0f64dcd96f53/F0.jpg", "media/e3c46a5a0f64dcd96f53/F1.jpg", "media/e3c46a5a0f64dcd96f53/F2.jpg", "media/e3c46a5a0f64dcd96f53/F3.jpg", "media/e3c46a5a0f64dcd96f53/F4.jpg", "media/e3c46a5a0f64dcd96f53/F5.jpg", "media/e3c46a5a0f64dcd96f53/F6.jpg", "media/e3c46a5a0f64dcd96f53/F7.jpg", "media/e3c46a5a0f64dcd96f53/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "490e4d66be7151a75fcabd7ff061a6b5c4c994f1b37afb2d7f6f01c408501b9c"}
{"qa_id": "VOST__1456_cut_carrot__object_pair_distance_change__01", "video_id": "VOST__1456_cut_carrot", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, did the 3D distance between object A (hand holding a knife) and object B (hand holding a carrot) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/7941f436ef7b31bbab55/F0.jpg", "media/7941f436ef7b31bbab55/F1.jpg", "media/7941f436ef7b31bbab55/F2.jpg", "media/7941f436ef7b31bbab55/F3.jpg", "media/7941f436ef7b31bbab55/F4.jpg", "media/7941f436ef7b31bbab55/F5.jpg", "media/7941f436ef7b31bbab55/F6.jpg", "media/7941f436ef7b31bbab55/F7.jpg", "media/7941f436ef7b31bbab55/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (hand holding a knife) and object B (hand holding a carrot) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d358c2cebd4faab62798f7e27d3571b7a799a6773036472286e864fd402c4c9d"}
{"qa_id": "VOST__1837_open_jar__camera_translation_direction__00", "video_id": "VOST__1837_open_jar", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/3909608264cd6ad3bb64/F0.jpg", "media/3909608264cd6ad3bb64/F1.jpg", "media/3909608264cd6ad3bb64/F2.jpg", "media/3909608264cd6ad3bb64/F3.jpg", "media/3909608264cd6ad3bb64/F4.jpg", "media/3909608264cd6ad3bb64/F5.jpg", "media/3909608264cd6ad3bb64/F6.jpg", "media/3909608264cd6ad3bb64/F7.jpg", "media/3909608264cd6ad3bb64/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "894ef8f27063bc6435e373fcbd408fa56d96ae266e50bceb99cfa10d94604496"}
{"qa_id": "VOST__1896_crush_box__camera_view_direction_change__00", "video_id": "VOST__1896_crush_box", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/03c84f4fde9eb8d051bf/F0.jpg", "media/03c84f4fde9eb8d051bf/F1.jpg", "media/03c84f4fde9eb8d051bf/F2.jpg", "media/03c84f4fde9eb8d051bf/F3.jpg", "media/03c84f4fde9eb8d051bf/F4.jpg", "media/03c84f4fde9eb8d051bf/F5.jpg", "media/03c84f4fde9eb8d051bf/F6.jpg", "media/03c84f4fde9eb8d051bf/F7.jpg", "media/03c84f4fde9eb8d051bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88ac3e3ed5ed47bb46dab59f07f16f30445545b8c4453afd953d5bb8abdbf9d7"}
{"qa_id": "VOST__1896_crush_box__object_travel_distance_comparison__01", "video_id": "VOST__1896_crush_box", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm wearing a blue long-sleeved shirt and a silver watch) or B (right arm wearing a blue long-sleeved shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/8f0e4261a35796a45aa9/F0.jpg", "media/8f0e4261a35796a45aa9/F1.jpg", "media/8f0e4261a35796a45aa9/F2.jpg", "media/8f0e4261a35796a45aa9/F3.jpg", "media/8f0e4261a35796a45aa9/F4.jpg", "media/8f0e4261a35796a45aa9/F5.jpg", "media/8f0e4261a35796a45aa9/F6.jpg", "media/8f0e4261a35796a45aa9/F7.jpg", "media/8f0e4261a35796a45aa9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm wearing a blue long-sleeved shirt and a silver watch) or B (right arm wearing a blue long-sleeved shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7ad589f0948642eb05beaa7cd1b0992b06d3a42bfebe88f8212719df642899e"}
{"qa_id": "VOST__2539_mix_biscuit__camera_translation_direction__00", "video_id": "VOST__2539_mix_biscuit", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "forward", "up"], "frame_paths": ["media/7d306b0c561215c74df4/F0.jpg", "media/7d306b0c561215c74df4/F1.jpg", "media/7d306b0c561215c74df4/F2.jpg", "media/7d306b0c561215c74df4/F3.jpg", "media/7d306b0c561215c74df4/F4.jpg", "media/7d306b0c561215c74df4/F5.jpg", "media/7d306b0c561215c74df4/F6.jpg", "media/7d306b0c561215c74df4/F7.jpg", "media/7d306b0c561215c74df4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a1851e02bf17e67acb125a09af36df1d91c2834f52ad35e08a153b0c9fec8fa6"}
{"qa_id": "VOST__2596_mix_dough__camera_translation_direction__00", "video_id": "VOST__2596_mix_dough", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "left", "forward"], "frame_paths": ["media/1fda546f978c1cd463fd/F0.jpg", "media/1fda546f978c1cd463fd/F1.jpg", "media/1fda546f978c1cd463fd/F2.jpg", "media/1fda546f978c1cd463fd/F3.jpg", "media/1fda546f978c1cd463fd/F4.jpg", "media/1fda546f978c1cd463fd/F5.jpg", "media/1fda546f978c1cd463fd/F6.jpg", "media/1fda546f978c1cd463fd/F7.jpg", "media/1fda546f978c1cd463fd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c8df04d54e586e877738c941ae2053294ce05c225f76769f76720a3976d4f2e"}
{"qa_id": "VOST__2731_scoop_vegetable__camera_translation_direction__00", "video_id": "VOST__2731_scoop_vegetable", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "up", "left", "backward", "forward"], "frame_paths": ["media/12b1462aa4ea086fb7d8/F0.jpg", "media/12b1462aa4ea086fb7d8/F1.jpg", "media/12b1462aa4ea086fb7d8/F2.jpg", "media/12b1462aa4ea086fb7d8/F3.jpg", "media/12b1462aa4ea086fb7d8/F4.jpg", "media/12b1462aa4ea086fb7d8/F5.jpg", "media/12b1462aa4ea086fb7d8/F6.jpg", "media/12b1462aa4ea086fb7d8/F7.jpg", "media/12b1462aa4ea086fb7d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. up\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e168522f87a581ae08dd32012b81f2d01e24d2ec00d7925f89846d125b660c3"}
{"qa_id": "VOST__2731_scoop_vegetable__object_world_motion_direction__01", "video_id": "VOST__2731_scoop_vegetable", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (hand holding a utensil) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "backward", "right", "up"], "frame_paths": ["media/db19bc324d59296044e4/F0.jpg", "media/db19bc324d59296044e4/F1.jpg", "media/db19bc324d59296044e4/F2.jpg", "media/db19bc324d59296044e4/F3.jpg", "media/db19bc324d59296044e4/F4.jpg", "media/db19bc324d59296044e4/F5.jpg", "media/db19bc324d59296044e4/F6.jpg", "media/db19bc324d59296044e4/F7.jpg", "media/db19bc324d59296044e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (hand holding a utensil) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9da2b56844c25c03acee458d9903461c84e725822b496a335722a2af1c11f705"}
{"qa_id": "VOST__2731_scoop_vegetable__object_world_motion_direction__02", "video_id": "VOST__2731_scoop_vegetable", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object B (metal colander containing chopped green vegetables) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "up", "down", "right"], "frame_paths": ["media/99b91bf9f013919058a8/F0.jpg", "media/99b91bf9f013919058a8/F1.jpg", "media/99b91bf9f013919058a8/F2.jpg", "media/99b91bf9f013919058a8/F3.jpg", "media/99b91bf9f013919058a8/F4.jpg", "media/99b91bf9f013919058a8/F5.jpg", "media/99b91bf9f013919058a8/F6.jpg", "media/99b91bf9f013919058a8/F7.jpg", "media/99b91bf9f013919058a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object B (metal colander containing chopped green vegetables) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "397163df6914c3b364e04a03310e92c0ff7a144de4b27cae64159310aa324d86"}
{"qa_id": "VOST__2731_scoop_vegetable__object_travel_distance_comparison__03", "video_id": "VOST__2731_scoop_vegetable", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (hand holding a utensil) or B (metal colander containing chopped green vegetables)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/83158a0fab7e2f946d95/F0.jpg", "media/83158a0fab7e2f946d95/F1.jpg", "media/83158a0fab7e2f946d95/F2.jpg", "media/83158a0fab7e2f946d95/F3.jpg", "media/83158a0fab7e2f946d95/F4.jpg", "media/83158a0fab7e2f946d95/F5.jpg", "media/83158a0fab7e2f946d95/F6.jpg", "media/83158a0fab7e2f946d95/F7.jpg", "media/83158a0fab7e2f946d95/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (hand holding a utensil) or B (metal colander containing chopped green vegetables)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9cf2513cac57a844b476baf7a52fd6ab4ab620590cca2aa986a84718123483e3"}
{"qa_id": "VOST__2765_scoop_jam__camera_translation_direction__00", "video_id": "VOST__2765_scoop_jam", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "forward", "left", "right"], "frame_paths": ["media/8768b661189d5624f2b6/F0.jpg", "media/8768b661189d5624f2b6/F1.jpg", "media/8768b661189d5624f2b6/F2.jpg", "media/8768b661189d5624f2b6/F3.jpg", "media/8768b661189d5624f2b6/F4.jpg", "media/8768b661189d5624f2b6/F5.jpg", "media/8768b661189d5624f2b6/F6.jpg", "media/8768b661189d5624f2b6/F7.jpg", "media/8768b661189d5624f2b6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e168b909f8085a54a4ba9059e37bb8f75c4d5865712cc55f6b260601cc52569"}
{"qa_id": "VOST__2788_scoop_jam__camera_translation_direction__00", "video_id": "VOST__2788_scoop_jam", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "left", "forward", "backward", "down"], "frame_paths": ["media/2b8b3ed2280bd4d91a84/F0.jpg", "media/2b8b3ed2280bd4d91a84/F1.jpg", "media/2b8b3ed2280bd4d91a84/F2.jpg", "media/2b8b3ed2280bd4d91a84/F3.jpg", "media/2b8b3ed2280bd4d91a84/F4.jpg", "media/2b8b3ed2280bd4d91a84/F5.jpg", "media/2b8b3ed2280bd4d91a84/F6.jpg", "media/2b8b3ed2280bd4d91a84/F7.jpg", "media/2b8b3ed2280bd4d91a84/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. left\nD. forward\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb5bf89851628be8f73ca11f1b00f07173ce7a5f6a9bb30edfc44ac93939557f"}
{"qa_id": "VOST__2788_scoop_jam__object_world_motion_direction__02", "video_id": "VOST__2788_scoop_jam", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object B (orange food substance on a slice of bread) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "up", "right", "forward"], "frame_paths": ["media/6c475a525cdc89016eaa/F0.jpg", "media/6c475a525cdc89016eaa/F1.jpg", "media/6c475a525cdc89016eaa/F2.jpg", "media/6c475a525cdc89016eaa/F3.jpg", "media/6c475a525cdc89016eaa/F4.jpg", "media/6c475a525cdc89016eaa/F5.jpg", "media/6c475a525cdc89016eaa/F6.jpg", "media/6c475a525cdc89016eaa/F7.jpg", "media/6c475a525cdc89016eaa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object B (orange food substance on a slice of bread) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f0c728e288d809456e5af118090e6ff29bb57a918eac2f73e1907059b1afcd9"}
{"qa_id": "VOST__283_squeeze_lime__camera_translation_direction__00", "video_id": "VOST__283_squeeze_lime", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "down", "forward"], "frame_paths": ["media/d9cbce2c79c3ab3e157e/F0.jpg", "media/d9cbce2c79c3ab3e157e/F1.jpg", "media/d9cbce2c79c3ab3e157e/F2.jpg", "media/d9cbce2c79c3ab3e157e/F3.jpg", "media/d9cbce2c79c3ab3e157e/F4.jpg", "media/d9cbce2c79c3ab3e157e/F5.jpg", "media/d9cbce2c79c3ab3e157e/F6.jpg", "media/d9cbce2c79c3ab3e157e/F7.jpg", "media/d9cbce2c79c3ab3e157e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "870557531b09741a15b9ed733f1a904dfb01265d37de5899ec3da049a85ce5d8"}
{"qa_id": "VOST__2840_scoop_food__camera_translation_direction__00", "video_id": "VOST__2840_scoop_food", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/85ba7ff0819bed022c0a/F0.jpg", "media/85ba7ff0819bed022c0a/F1.jpg", "media/85ba7ff0819bed022c0a/F2.jpg", "media/85ba7ff0819bed022c0a/F3.jpg", "media/85ba7ff0819bed022c0a/F4.jpg", "media/85ba7ff0819bed022c0a/F5.jpg", "media/85ba7ff0819bed022c0a/F6.jpg", "media/85ba7ff0819bed022c0a/F7.jpg", "media/85ba7ff0819bed022c0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c0caca1ea4bbe09f1169d009df3981800f048361d2928d86659b92f266626228"}
{"qa_id": "VOST__3131_scrape_stirfry__object_travel_distance_comparison__00", "video_id": "VOST__3131_scrape_stirfry", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm with a black watch) or B (right arm and lower body in dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/a03f787a4a33640bcd96/F0.jpg", "media/a03f787a4a33640bcd96/F1.jpg", "media/a03f787a4a33640bcd96/F2.jpg", "media/a03f787a4a33640bcd96/F3.jpg", "media/a03f787a4a33640bcd96/F4.jpg", "media/a03f787a4a33640bcd96/F5.jpg", "media/a03f787a4a33640bcd96/F6.jpg", "media/a03f787a4a33640bcd96/F7.jpg", "media/a03f787a4a33640bcd96/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm with a black watch) or B (right arm and lower body in dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36e878f7ac9124bb723e1c3c395e8cc217bb4110fe72296ff59b720e96b42831"}
{"qa_id": "VOST__3135_scrape_pan__object_pair_distance_change__00", "video_id": "VOST__3135_scrape_pan", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (person's arm and hand reaching to spread food) and object B (black frying pan containing cooked meat) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/7aa2e3213288d7c6f2da/F0.jpg", "media/7aa2e3213288d7c6f2da/F1.jpg", "media/7aa2e3213288d7c6f2da/F2.jpg", "media/7aa2e3213288d7c6f2da/F3.jpg", "media/7aa2e3213288d7c6f2da/F4.jpg", "media/7aa2e3213288d7c6f2da/F5.jpg", "media/7aa2e3213288d7c6f2da/F6.jpg", "media/7aa2e3213288d7c6f2da/F7.jpg", "media/7aa2e3213288d7c6f2da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person's arm and hand reaching to spread food) and object B (black frying pan containing cooked meat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1e18323beb9e6cdaff1cd5d83df48416e746f46d41c662e7a85798fb2e5ef9d0"}
{"qa_id": "VOST__3150_peel_fish__object_travel_distance_comparison__00", "video_id": "VOST__3150_peel_fish", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand) or B (black shirt with white floral pattern)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/a7ca5e810fb29e444d72/F0.jpg", "media/a7ca5e810fb29e444d72/F1.jpg", "media/a7ca5e810fb29e444d72/F2.jpg", "media/a7ca5e810fb29e444d72/F3.jpg", "media/a7ca5e810fb29e444d72/F4.jpg", "media/a7ca5e810fb29e444d72/F5.jpg", "media/a7ca5e810fb29e444d72/F6.jpg", "media/a7ca5e810fb29e444d72/F7.jpg", "media/a7ca5e810fb29e444d72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand) or B (black shirt with white floral pattern)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e76eca41a831eab65df249a20d08c5e58fed2cb8ac72db04aab731b57e0fc183"}
{"qa_id": "VOST__3160_peel_banana__camera_translation_direction__00", "video_id": "VOST__3160_peel_banana", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "right", "down", "backward", "forward"], "frame_paths": ["media/b5af012ba98c0ee7875d/F0.jpg", "media/b5af012ba98c0ee7875d/F1.jpg", "media/b5af012ba98c0ee7875d/F2.jpg", "media/b5af012ba98c0ee7875d/F3.jpg", "media/b5af012ba98c0ee7875d/F4.jpg", "media/b5af012ba98c0ee7875d/F5.jpg", "media/b5af012ba98c0ee7875d/F6.jpg", "media/b5af012ba98c0ee7875d/F7.jpg", "media/b5af012ba98c0ee7875d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. right\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "23ca0d1b98a878896cf704d3951900b69eae4b15d5ec3bd08ff5afd79d823d0d"}
{"qa_id": "VOST__3177_peel_peach__camera_translation_direction__00", "video_id": "VOST__3177_peel_peach", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/004aef8c9d64d6f8331f/F0.jpg", "media/004aef8c9d64d6f8331f/F1.jpg", "media/004aef8c9d64d6f8331f/F2.jpg", "media/004aef8c9d64d6f8331f/F3.jpg", "media/004aef8c9d64d6f8331f/F4.jpg", "media/004aef8c9d64d6f8331f/F5.jpg", "media/004aef8c9d64d6f8331f/F6.jpg", "media/004aef8c9d64d6f8331f/F7.jpg", "media/004aef8c9d64d6f8331f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d4ff7cda1a61b3426015527252c0166a9dd01115a55d24363c9f120a5ce90d66"}
{"qa_id": "VOST__3292_peel_garlic__camera_translation_direction__00", "video_id": "VOST__3292_peel_garlic", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "forward", "down", "left"], "frame_paths": ["media/e5ddd63ed3e3e4867e0c/F0.jpg", "media/e5ddd63ed3e3e4867e0c/F1.jpg", "media/e5ddd63ed3e3e4867e0c/F2.jpg", "media/e5ddd63ed3e3e4867e0c/F3.jpg", "media/e5ddd63ed3e3e4867e0c/F4.jpg", "media/e5ddd63ed3e3e4867e0c/F5.jpg", "media/e5ddd63ed3e3e4867e0c/F6.jpg", "media/e5ddd63ed3e3e4867e0c/F7.jpg", "media/e5ddd63ed3e3e4867e0c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8666e7a7541f1c8b1603790e5ea31a36bfc195efa8680093a28dd835cbc1d01d"}
{"qa_id": "VOST__3295_break_garlic__camera_translation_direction__00", "video_id": "VOST__3295_break_garlic", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "down", "left", "up", "right"], "frame_paths": ["media/ff638e886dde9a86d4d1/F0.jpg", "media/ff638e886dde9a86d4d1/F1.jpg", "media/ff638e886dde9a86d4d1/F2.jpg", "media/ff638e886dde9a86d4d1/F3.jpg", "media/ff638e886dde9a86d4d1/F4.jpg", "media/ff638e886dde9a86d4d1/F5.jpg", "media/ff638e886dde9a86d4d1/F6.jpg", "media/ff638e886dde9a86d4d1/F7.jpg", "media/ff638e886dde9a86d4d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. down\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e31551a3678388c404b28d80a847f1ae9b4f8107b17c76dea86ecb0b13265508"}
{"qa_id": "VOST__3303_open_salmon__camera_translation_direction__00", "video_id": "VOST__3303_open_salmon", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "up", "forward", "down"], "frame_paths": ["media/0693cc9c54e56c7190c3/F0.jpg", "media/0693cc9c54e56c7190c3/F1.jpg", "media/0693cc9c54e56c7190c3/F2.jpg", "media/0693cc9c54e56c7190c3/F3.jpg", "media/0693cc9c54e56c7190c3/F4.jpg", "media/0693cc9c54e56c7190c3/F5.jpg", "media/0693cc9c54e56c7190c3/F6.jpg", "media/0693cc9c54e56c7190c3/F7.jpg", "media/0693cc9c54e56c7190c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c7c3690a9d7a532de5e4a05b5f4e4a398806d052f4eca084f2017ad30fd846c"}
{"qa_id": "VOST__3303_open_salmon__camera_object_distance_change__01", "video_id": "VOST__3303_open_salmon", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, did object A (person wearing a dark green long-sleeved shirt with white stripes on the sleeve) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/66d97c4e3dcdc0434ad0/F0.jpg", "media/66d97c4e3dcdc0434ad0/F1.jpg", "media/66d97c4e3dcdc0434ad0/F2.jpg", "media/66d97c4e3dcdc0434ad0/F3.jpg", "media/66d97c4e3dcdc0434ad0/F4.jpg", "media/66d97c4e3dcdc0434ad0/F5.jpg", "media/66d97c4e3dcdc0434ad0/F6.jpg", "media/66d97c4e3dcdc0434ad0/F7.jpg", "media/66d97c4e3dcdc0434ad0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (person wearing a dark green long-sleeved shirt with white stripes on the sleeve) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "999773353c6914e3b69c1224751a0ea03f460be4da6803dba07f784bfa2dfd6a"}
{"qa_id": "VOST__3303_open_salmon__object_pair_distance_change__02", "video_id": "VOST__3303_open_salmon", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, did the 3D distance between object A (person wearing a dark green long-sleeved shirt with white stripes on the sleeve) and object B (chopped green vegetables in a white bowl with a blue rim) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/ffe1f3b4f7bb41852820/F0.jpg", "media/ffe1f3b4f7bb41852820/F1.jpg", "media/ffe1f3b4f7bb41852820/F2.jpg", "media/ffe1f3b4f7bb41852820/F3.jpg", "media/ffe1f3b4f7bb41852820/F4.jpg", "media/ffe1f3b4f7bb41852820/F5.jpg", "media/ffe1f3b4f7bb41852820/F6.jpg", "media/ffe1f3b4f7bb41852820/F7.jpg", "media/ffe1f3b4f7bb41852820/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did the 3D distance between object A (person wearing a dark green long-sleeved shirt with white stripes on the sleeve) and object B (chopped green vegetables in a white bowl with a blue rim) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bce88b72f75e1539825d167a6fa5ab8a0bf39a1d0bc13f7b9286ec9f371ebfbb"}
{"qa_id": "VOST__3304_open_package__camera_translation_direction__00", "video_id": "VOST__3304_open_package", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "forward", "backward", "down"], "frame_paths": ["media/79624f65212bdd9e5746/F0.jpg", "media/79624f65212bdd9e5746/F1.jpg", "media/79624f65212bdd9e5746/F2.jpg", "media/79624f65212bdd9e5746/F3.jpg", "media/79624f65212bdd9e5746/F4.jpg", "media/79624f65212bdd9e5746/F5.jpg", "media/79624f65212bdd9e5746/F6.jpg", "media/79624f65212bdd9e5746/F7.jpg", "media/79624f65212bdd9e5746/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. forward\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c96ee2f4fe94b1e584c727c61995a25d11a22663a70dff5468f87337bdded3a7"}
{"qa_id": "VOST__3304_open_package__object_world_motion_direction__01", "video_id": "VOST__3304_open_package", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F7, in which dominant direction did object B (person's left arm and hand) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/e19f2da14ed1cba8235d/F0.jpg", "media/e19f2da14ed1cba8235d/F1.jpg", "media/e19f2da14ed1cba8235d/F2.jpg", "media/e19f2da14ed1cba8235d/F3.jpg", "media/e19f2da14ed1cba8235d/F4.jpg", "media/e19f2da14ed1cba8235d/F5.jpg", "media/e19f2da14ed1cba8235d/F6.jpg", "media/e19f2da14ed1cba8235d/F7.jpg", "media/e19f2da14ed1cba8235d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F7, in which dominant direction did object B (person's left arm and hand) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f644015a42172662201a1a4863133a685a5897491a872470338588c945a0fddf"}
{"qa_id": "VOST__3304_open_package__object_pair_distance_change__02", "video_id": "VOST__3304_open_package", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, did the 3D distance between object A (person's right arm and hand) and object B (person's left arm and hand) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/aeb2d642e04a35845c72/F0.jpg", "media/aeb2d642e04a35845c72/F1.jpg", "media/aeb2d642e04a35845c72/F2.jpg", "media/aeb2d642e04a35845c72/F3.jpg", "media/aeb2d642e04a35845c72/F4.jpg", "media/aeb2d642e04a35845c72/F5.jpg", "media/aeb2d642e04a35845c72/F6.jpg", "media/aeb2d642e04a35845c72/F7.jpg", "media/aeb2d642e04a35845c72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did the 3D distance between object A (person's right arm and hand) and object B (person's left arm and hand) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "20a7d9b44c52c03f6512db5b4afce678397711a83f5a01a6416ef50470867c18"}
{"qa_id": "VOST__3304_open_package__object_travel_distance_comparison__03", "video_id": "VOST__3304_open_package", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand) or B (person's left arm and hand)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/aeb2d642e04a35845c72/F0.jpg", "media/aeb2d642e04a35845c72/F1.jpg", "media/aeb2d642e04a35845c72/F2.jpg", "media/aeb2d642e04a35845c72/F3.jpg", "media/aeb2d642e04a35845c72/F4.jpg", "media/aeb2d642e04a35845c72/F5.jpg", "media/aeb2d642e04a35845c72/F6.jpg", "media/aeb2d642e04a35845c72/F7.jpg", "media/aeb2d642e04a35845c72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand) or B (person's left arm and hand)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65385c61b0497030fd3d168fd336b05b20630316fa2453b5959d45804321a358"}
{"qa_id": "VOST__3305_peel_fish__camera_translation_direction__00", "video_id": "VOST__3305_peel_fish", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "up", "right", "left"], "frame_paths": ["media/16eb1029b5b345fa6c1c/F0.jpg", "media/16eb1029b5b345fa6c1c/F1.jpg", "media/16eb1029b5b345fa6c1c/F2.jpg", "media/16eb1029b5b345fa6c1c/F3.jpg", "media/16eb1029b5b345fa6c1c/F4.jpg", "media/16eb1029b5b345fa6c1c/F5.jpg", "media/16eb1029b5b345fa6c1c/F6.jpg", "media/16eb1029b5b345fa6c1c/F7.jpg", "media/16eb1029b5b345fa6c1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. up\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b6e323f62f8fd5eec62a62da0677cb35a6359a33714a759e38e93cb0364d21ab"}
{"qa_id": "VOST__3305_peel_fish__object_pair_distance_change__01", "video_id": "VOST__3305_peel_fish", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, did the 3D distance between object A (right arm of a person wearing a grey long-sleeved shirt) and object B (left arm of a person wearing a dark long-sleeved shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/1b769943482295f6ed95/F0.jpg", "media/1b769943482295f6ed95/F1.jpg", "media/1b769943482295f6ed95/F2.jpg", "media/1b769943482295f6ed95/F3.jpg", "media/1b769943482295f6ed95/F4.jpg", "media/1b769943482295f6ed95/F5.jpg", "media/1b769943482295f6ed95/F6.jpg", "media/1b769943482295f6ed95/F7.jpg", "media/1b769943482295f6ed95/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did the 3D distance between object A (right arm of a person wearing a grey long-sleeved shirt) and object B (left arm of a person wearing a dark long-sleeved shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b293b0166451e6b8681d395ea6403e52950ba3ef340333742ccdea88fa4b8cc5"}
{"qa_id": "VOST__3317_peel_onion__camera_translation_direction__00", "video_id": "VOST__3317_peel_onion", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "down", "forward", "right"], "frame_paths": ["media/a51b5479e99571c6cd26/F0.jpg", "media/a51b5479e99571c6cd26/F1.jpg", "media/a51b5479e99571c6cd26/F2.jpg", "media/a51b5479e99571c6cd26/F3.jpg", "media/a51b5479e99571c6cd26/F4.jpg", "media/a51b5479e99571c6cd26/F5.jpg", "media/a51b5479e99571c6cd26/F6.jpg", "media/a51b5479e99571c6cd26/F7.jpg", "media/a51b5479e99571c6cd26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62a84cbf6ed43c7d1201e5e925b5a906224f15a2afd1f2f81028911fdbe2e50a"}
{"qa_id": "VOST__3317_peel_onion__camera_view_direction_change__01", "video_id": "VOST__3317_peel_onion", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/a51b5479e99571c6cd26/F0.jpg", "media/a51b5479e99571c6cd26/F1.jpg", "media/a51b5479e99571c6cd26/F2.jpg", "media/a51b5479e99571c6cd26/F3.jpg", "media/a51b5479e99571c6cd26/F4.jpg", "media/a51b5479e99571c6cd26/F5.jpg", "media/a51b5479e99571c6cd26/F6.jpg", "media/a51b5479e99571c6cd26/F7.jpg", "media/a51b5479e99571c6cd26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "603de446c5814db82051c70f8311b3a130eb38a2dbf02023686fab05c41c4c76"}
{"qa_id": "VOST__3318_peel_potato__camera_translation_direction__00", "video_id": "VOST__3318_peel_potato", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "up", "left", "backward", "down"], "frame_paths": ["media/7c9a05b845a482eb46f9/F0.jpg", "media/7c9a05b845a482eb46f9/F1.jpg", "media/7c9a05b845a482eb46f9/F2.jpg", "media/7c9a05b845a482eb46f9/F3.jpg", "media/7c9a05b845a482eb46f9/F4.jpg", "media/7c9a05b845a482eb46f9/F5.jpg", "media/7c9a05b845a482eb46f9/F6.jpg", "media/7c9a05b845a482eb46f9/F7.jpg", "media/7c9a05b845a482eb46f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. up\nD. left\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f163ab5caebe1f5d06fc5b21e4d9bd9ac694b8dbae4dbd8fe90ea6d34819275c"}
{"qa_id": "VOST__3318_peel_potato__object_world_motion_direction__01", "video_id": "VOST__3318_peel_potato", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a black shirt, visible from the chest up, with arms extended towards a cutting board) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "left", "down", "forward", "up"], "frame_paths": ["media/619cb5eed0c4bd136c56/F0.jpg", "media/619cb5eed0c4bd136c56/F1.jpg", "media/619cb5eed0c4bd136c56/F2.jpg", "media/619cb5eed0c4bd136c56/F3.jpg", "media/619cb5eed0c4bd136c56/F4.jpg", "media/619cb5eed0c4bd136c56/F5.jpg", "media/619cb5eed0c4bd136c56/F6.jpg", "media/619cb5eed0c4bd136c56/F7.jpg", "media/619cb5eed0c4bd136c56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a black shirt, visible from the chest up, with arms extended towards a cutting board) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. left\nD. down\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d3d93f36bf6e0b3dbae8a1152c5d1123598a490cd43a23954abfc7d5ccf14b3"}
{"qa_id": "VOST__3321_peel_onion__camera_translation_direction__00", "video_id": "VOST__3321_peel_onion", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "right", "down"], "frame_paths": ["media/1630e2e21f77a9e3207d/F0.jpg", "media/1630e2e21f77a9e3207d/F1.jpg", "media/1630e2e21f77a9e3207d/F2.jpg", "media/1630e2e21f77a9e3207d/F3.jpg", "media/1630e2e21f77a9e3207d/F4.jpg", "media/1630e2e21f77a9e3207d/F5.jpg", "media/1630e2e21f77a9e3207d/F6.jpg", "media/1630e2e21f77a9e3207d/F7.jpg", "media/1630e2e21f77a9e3207d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a8aa03e20f7a46d2817b103ea150547b216333fec21d63b465f9eff4e363b6e8"}
{"qa_id": "VOST__3322_peel_potato__camera_translation_direction__00", "video_id": "VOST__3322_peel_potato", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "right", "left", "backward"], "frame_paths": ["media/b6fda15f19b8c7919504/F0.jpg", "media/b6fda15f19b8c7919504/F1.jpg", "media/b6fda15f19b8c7919504/F2.jpg", "media/b6fda15f19b8c7919504/F3.jpg", "media/b6fda15f19b8c7919504/F4.jpg", "media/b6fda15f19b8c7919504/F5.jpg", "media/b6fda15f19b8c7919504/F6.jpg", "media/b6fda15f19b8c7919504/F7.jpg", "media/b6fda15f19b8c7919504/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. right\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0b2487de91169258b32ee22cedf356c8f3b7a0b9dede4a92dcdade39f098165"}
{"qa_id": "VOST__3322_peel_potato__object_world_motion_direction__01", "video_id": "VOST__3322_peel_potato", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (yellow fruit, likely a banana, being peeled) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "backward", "down", "left"], "frame_paths": ["media/a5052887da3d65f91d66/F0.jpg", "media/a5052887da3d65f91d66/F1.jpg", "media/a5052887da3d65f91d66/F2.jpg", "media/a5052887da3d65f91d66/F3.jpg", "media/a5052887da3d65f91d66/F4.jpg", "media/a5052887da3d65f91d66/F5.jpg", "media/a5052887da3d65f91d66/F6.jpg", "media/a5052887da3d65f91d66/F7.jpg", "media/a5052887da3d65f91d66/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (yellow fruit, likely a banana, being peeled) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d78ce0bc12850618fc773e52aa5df7ed76fc417fe984d2f2f23746bdb13035bb"}
{"qa_id": "VOST__3345_peel_courgette__camera_translation_direction__00", "video_id": "VOST__3345_peel_courgette", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "right", "backward", "up"], "frame_paths": ["media/a3a63e8214992a1dd6d1/F0.jpg", "media/a3a63e8214992a1dd6d1/F1.jpg", "media/a3a63e8214992a1dd6d1/F2.jpg", "media/a3a63e8214992a1dd6d1/F3.jpg", "media/a3a63e8214992a1dd6d1/F4.jpg", "media/a3a63e8214992a1dd6d1/F5.jpg", "media/a3a63e8214992a1dd6d1/F6.jpg", "media/a3a63e8214992a1dd6d1/F7.jpg", "media/a3a63e8214992a1dd6d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "25b05424061cfd5717d4fcc5ac880ed457b3f80cc06532ac0b94c3476dd674e6"}
{"qa_id": "VOST__3447_squeeze_sausage__object_travel_distance_comparison__00", "video_id": "VOST__3447_squeeze_sausage", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm and hand) or B (right arm and hand with blue wristband)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/25793db48264f5adbd93/F0.jpg", "media/25793db48264f5adbd93/F1.jpg", "media/25793db48264f5adbd93/F2.jpg", "media/25793db48264f5adbd93/F3.jpg", "media/25793db48264f5adbd93/F4.jpg", "media/25793db48264f5adbd93/F5.jpg", "media/25793db48264f5adbd93/F6.jpg", "media/25793db48264f5adbd93/F7.jpg", "media/25793db48264f5adbd93/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (left arm and hand) or B (right arm and hand with blue wristband)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9342ce8249d62b402bc50b8d2d61cd659b40bb4a943d76987db0a5df93b98d01"}
{"qa_id": "VOST__3496_peel_onion__camera_translation_direction__00", "video_id": "VOST__3496_peel_onion", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "forward", "left"], "frame_paths": ["media/6c2da5ad42e1d2d0a4d5/F0.jpg", "media/6c2da5ad42e1d2d0a4d5/F1.jpg", "media/6c2da5ad42e1d2d0a4d5/F2.jpg", "media/6c2da5ad42e1d2d0a4d5/F3.jpg", "media/6c2da5ad42e1d2d0a4d5/F4.jpg", "media/6c2da5ad42e1d2d0a4d5/F5.jpg", "media/6c2da5ad42e1d2d0a4d5/F6.jpg", "media/6c2da5ad42e1d2d0a4d5/F7.jpg", "media/6c2da5ad42e1d2d0a4d5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "115c892c6fadcaf379d50e3cca6b4ebc3e8e4e2edca16e09f002ea32a43be087"}
{"qa_id": "VOST__3687_rip_foil__camera_translation_direction__00", "video_id": "VOST__3687_rip_foil", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "forward", "backward", "up"], "frame_paths": ["media/25a590f0e15a60a07fba/F0.jpg", "media/25a590f0e15a60a07fba/F1.jpg", "media/25a590f0e15a60a07fba/F2.jpg", "media/25a590f0e15a60a07fba/F3.jpg", "media/25a590f0e15a60a07fba/F4.jpg", "media/25a590f0e15a60a07fba/F5.jpg", "media/25a590f0e15a60a07fba/F6.jpg", "media/25a590f0e15a60a07fba/F7.jpg", "media/25a590f0e15a60a07fba/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9e2d4b6b2f5585130bc7b7f819b2322f715b4f030e73a8eb6bec767040388232"}
{"qa_id": "VOST__3687_rip_foil__object_world_motion_direction__01", "video_id": "VOST__3687_rip_foil", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (rectangular package wrapped in clear plastic with a purple label) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "right", "left", "backward"], "frame_paths": ["media/17e686a23dfaaa1ed403/F0.jpg", "media/17e686a23dfaaa1ed403/F1.jpg", "media/17e686a23dfaaa1ed403/F2.jpg", "media/17e686a23dfaaa1ed403/F3.jpg", "media/17e686a23dfaaa1ed403/F4.jpg", "media/17e686a23dfaaa1ed403/F5.jpg", "media/17e686a23dfaaa1ed403/F6.jpg", "media/17e686a23dfaaa1ed403/F7.jpg", "media/17e686a23dfaaa1ed403/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (rectangular package wrapped in clear plastic with a purple label) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. right\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "95a66d7d48c6db070489263bd353213e1ccfbdd90d3c972d2fb0e03c4a43166e"}
{"qa_id": "VOST__3702_grate_cheese__camera_translation_direction__00", "video_id": "VOST__3702_grate_cheese", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "forward", "backward", "up"], "frame_paths": ["media/3458d93ce3d46cd6d889/F0.jpg", "media/3458d93ce3d46cd6d889/F1.jpg", "media/3458d93ce3d46cd6d889/F2.jpg", "media/3458d93ce3d46cd6d889/F3.jpg", "media/3458d93ce3d46cd6d889/F4.jpg", "media/3458d93ce3d46cd6d889/F5.jpg", "media/3458d93ce3d46cd6d889/F6.jpg", "media/3458d93ce3d46cd6d889/F7.jpg", "media/3458d93ce3d46cd6d889/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "272265b42a5df6201d009eeb316fdf09d0561d72e11894fea9799c2e2724b4fb"}
{"qa_id": "VOST__3702_grate_cheese__object_world_motion_direction__01", "video_id": "VOST__3702_grate_cheese", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (person's left arm and hand holding a grater) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "right", "left", "up"], "frame_paths": ["media/c2bdffc9b920ab501af8/F0.jpg", "media/c2bdffc9b920ab501af8/F1.jpg", "media/c2bdffc9b920ab501af8/F2.jpg", "media/c2bdffc9b920ab501af8/F3.jpg", "media/c2bdffc9b920ab501af8/F4.jpg", "media/c2bdffc9b920ab501af8/F5.jpg", "media/c2bdffc9b920ab501af8/F6.jpg", "media/c2bdffc9b920ab501af8/F7.jpg", "media/c2bdffc9b920ab501af8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (person's left arm and hand holding a grater) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f635ce95971a1353e10937fd706836af225c416d09a563c0086f6a0141b10e68"}
{"qa_id": "VOST__3823_cut_paper__camera_translation_direction__00", "video_id": "VOST__3823_cut_paper", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "forward", "backward", "up"], "frame_paths": ["media/df2e292c273a69ff8d34/F0.jpg", "media/df2e292c273a69ff8d34/F1.jpg", "media/df2e292c273a69ff8d34/F2.jpg", "media/df2e292c273a69ff8d34/F3.jpg", "media/df2e292c273a69ff8d34/F4.jpg", "media/df2e292c273a69ff8d34/F5.jpg", "media/df2e292c273a69ff8d34/F6.jpg", "media/df2e292c273a69ff8d34/F7.jpg", "media/df2e292c273a69ff8d34/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "564546e39238ed984b316c7c08f7b1e63a5a3aedfa00fda918b6f7463423bea9"}
{"qa_id": "VOST__3827_cut_paper__object_travel_distance_comparison__00", "video_id": "VOST__3827_cut_paper", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a tool) or B (person's left arm and hand)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/ab72b4b98e17f2c791dd/F0.jpg", "media/ab72b4b98e17f2c791dd/F1.jpg", "media/ab72b4b98e17f2c791dd/F2.jpg", "media/ab72b4b98e17f2c791dd/F3.jpg", "media/ab72b4b98e17f2c791dd/F4.jpg", "media/ab72b4b98e17f2c791dd/F5.jpg", "media/ab72b4b98e17f2c791dd/F6.jpg", "media/ab72b4b98e17f2c791dd/F7.jpg", "media/ab72b4b98e17f2c791dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a tool) or B (person's left arm and hand)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1485b1c4e9063290919a32df2cb44ad8a2d7f68399b60e94d8aa5abf3259add7"}
{"qa_id": "VOST__3840_divide_dough__camera_translation_direction__00", "video_id": "VOST__3840_divide_dough", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "right", "down", "left", "backward"], "frame_paths": ["media/b34bfe4f843c0f25eee4/F0.jpg", "media/b34bfe4f843c0f25eee4/F1.jpg", "media/b34bfe4f843c0f25eee4/F2.jpg", "media/b34bfe4f843c0f25eee4/F3.jpg", "media/b34bfe4f843c0f25eee4/F4.jpg", "media/b34bfe4f843c0f25eee4/F5.jpg", "media/b34bfe4f843c0f25eee4/F6.jpg", "media/b34bfe4f843c0f25eee4/F7.jpg", "media/b34bfe4f843c0f25eee4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. right\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c9c91dccc38cbd2ba8ad817e7e8e46ac9481ea05c28e94e18350f32c05b1a712"}
{"qa_id": "VOST__4016_cut_herbs__camera_translation_direction__00", "video_id": "VOST__4016_cut_herbs", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "right", "forward", "up", "left"], "frame_paths": ["media/da134ee25acfb4970b89/F0.jpg", "media/da134ee25acfb4970b89/F1.jpg", "media/da134ee25acfb4970b89/F2.jpg", "media/da134ee25acfb4970b89/F3.jpg", "media/da134ee25acfb4970b89/F4.jpg", "media/da134ee25acfb4970b89/F5.jpg", "media/da134ee25acfb4970b89/F6.jpg", "media/da134ee25acfb4970b89/F7.jpg", "media/da134ee25acfb4970b89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. right\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e16309939616b0f808a22c1ed96ed3a653df13e1e2a695cddde0b4a0cae1f21"}
{"qa_id": "VOST__4016_cut_herbs__camera_view_direction_change__01", "video_id": "VOST__4016_cut_herbs", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/da134ee25acfb4970b89/F0.jpg", "media/da134ee25acfb4970b89/F1.jpg", "media/da134ee25acfb4970b89/F2.jpg", "media/da134ee25acfb4970b89/F3.jpg", "media/da134ee25acfb4970b89/F4.jpg", "media/da134ee25acfb4970b89/F5.jpg", "media/da134ee25acfb4970b89/F6.jpg", "media/da134ee25acfb4970b89/F7.jpg", "media/da134ee25acfb4970b89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4383964f8b2e713f8ef6219470edd1e1a95e07e00f5f6f81fafa727de656a97"}
{"qa_id": "VOST__4053_saw_wood__object_pair_distance_change__00", "video_id": "VOST__4053_saw_wood", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did the 3D distance between object A (person wearing a black t-shirt and a black wristband) and object B (yellow and black miter saw) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/5a7e9cf9001a709bc439/F0.jpg", "media/5a7e9cf9001a709bc439/F1.jpg", "media/5a7e9cf9001a709bc439/F2.jpg", "media/5a7e9cf9001a709bc439/F3.jpg", "media/5a7e9cf9001a709bc439/F4.jpg", "media/5a7e9cf9001a709bc439/F5.jpg", "media/5a7e9cf9001a709bc439/F6.jpg", "media/5a7e9cf9001a709bc439/F7.jpg", "media/5a7e9cf9001a709bc439/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did the 3D distance between object A (person wearing a black t-shirt and a black wristband) and object B (yellow and black miter saw) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "810ab0c6d2ec66d08def445c8c86a91aefd9a23a8de27229cacd4cece8e348c7"}
{"qa_id": "VOST__4057_break_garlic__object_world_motion_direction__00", "video_id": "VOST__4057_break_garlic", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person's left arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "up", "down", "backward"], "frame_paths": ["media/86488b3cf4f4a4cbb226/F0.jpg", "media/86488b3cf4f4a4cbb226/F1.jpg", "media/86488b3cf4f4a4cbb226/F2.jpg", "media/86488b3cf4f4a4cbb226/F3.jpg", "media/86488b3cf4f4a4cbb226/F4.jpg", "media/86488b3cf4f4a4cbb226/F5.jpg", "media/86488b3cf4f4a4cbb226/F6.jpg", "media/86488b3cf4f4a4cbb226/F7.jpg", "media/86488b3cf4f4a4cbb226/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person's left arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "288d6068df104b8dcbde63e8c5eafade63ad47a41f8aadd4cd53b9bc2ac154ef"}
{"qa_id": "VOST__4057_break_garlic__object_travel_distance_comparison__01", "video_id": "VOST__4057_break_garlic", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board) or B (person's right arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1c16473fab1550652a20/F0.jpg", "media/1c16473fab1550652a20/F1.jpg", "media/1c16473fab1550652a20/F2.jpg", "media/1c16473fab1550652a20/F3.jpg", "media/1c16473fab1550652a20/F4.jpg", "media/1c16473fab1550652a20/F5.jpg", "media/1c16473fab1550652a20/F6.jpg", "media/1c16473fab1550652a20/F7.jpg", "media/1c16473fab1550652a20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board) or B (person's right arm and hand wearing a dark long-sleeved shirt with a grey cuff, reaching towards a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0ea4bd9b48989143bcc4a3b1949bbfa78f560c0ec158b12b134b7b2ce1880b22"}
{"qa_id": "VOST__4079_cut_pea__object_pair_distance_change__00", "video_id": "VOST__4079_cut_pea", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, did the 3D distance between object A (right hand holding a knife) and object B (left hand holding vegetables) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/c628020994c16ed0d604/F0.jpg", "media/c628020994c16ed0d604/F1.jpg", "media/c628020994c16ed0d604/F2.jpg", "media/c628020994c16ed0d604/F3.jpg", "media/c628020994c16ed0d604/F4.jpg", "media/c628020994c16ed0d604/F5.jpg", "media/c628020994c16ed0d604/F6.jpg", "media/c628020994c16ed0d604/F7.jpg", "media/c628020994c16ed0d604/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did the 3D distance between object A (right hand holding a knife) and object B (left hand holding vegetables) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b86c5858f9c222d9bbc08207b90f68780fb92dec1aa733d3f63a412fa34aa79"}
{"qa_id": "VOST__4079_cut_pea__object_travel_distance_comparison__01", "video_id": "VOST__4079_cut_pea", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding vegetables)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/c628020994c16ed0d604/F0.jpg", "media/c628020994c16ed0d604/F1.jpg", "media/c628020994c16ed0d604/F2.jpg", "media/c628020994c16ed0d604/F3.jpg", "media/c628020994c16ed0d604/F4.jpg", "media/c628020994c16ed0d604/F5.jpg", "media/c628020994c16ed0d604/F6.jpg", "media/c628020994c16ed0d604/F7.jpg", "media/c628020994c16ed0d604/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding vegetables)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db88ef7bb68d7e7c42538c540a255eefc71411f913fdfdb1f8e203643a30dc1d"}
{"qa_id": "VOST__4174_cut_cloth__camera_translation_direction__00", "video_id": "VOST__4174_cut_cloth", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/0e701665fc5518f6e278/F0.jpg", "media/0e701665fc5518f6e278/F1.jpg", "media/0e701665fc5518f6e278/F2.jpg", "media/0e701665fc5518f6e278/F3.jpg", "media/0e701665fc5518f6e278/F4.jpg", "media/0e701665fc5518f6e278/F5.jpg", "media/0e701665fc5518f6e278/F6.jpg", "media/0e701665fc5518f6e278/F7.jpg", "media/0e701665fc5518f6e278/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0856b555dfe23de264caaf237019f8ccb1ca552c3801b0ee2b7a8f0cf2d7eaee"}
{"qa_id": "VOST__4176_cut_cloth__camera_translation_direction__00", "video_id": "VOST__4176_cut_cloth", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "right", "down", "left"], "frame_paths": ["media/e83ebb4ae76354d99cd9/F0.jpg", "media/e83ebb4ae76354d99cd9/F1.jpg", "media/e83ebb4ae76354d99cd9/F2.jpg", "media/e83ebb4ae76354d99cd9/F3.jpg", "media/e83ebb4ae76354d99cd9/F4.jpg", "media/e83ebb4ae76354d99cd9/F5.jpg", "media/e83ebb4ae76354d99cd9/F6.jpg", "media/e83ebb4ae76354d99cd9/F7.jpg", "media/e83ebb4ae76354d99cd9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f3a1fd3827dcad17afef8943e4e32a43bd08f4c4d81ab9ab361832de1cb29e5"}
{"qa_id": "VOST__4176_cut_cloth__camera_view_direction_change__01", "video_id": "VOST__4176_cut_cloth", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/e83ebb4ae76354d99cd9/F0.jpg", "media/e83ebb4ae76354d99cd9/F1.jpg", "media/e83ebb4ae76354d99cd9/F2.jpg", "media/e83ebb4ae76354d99cd9/F3.jpg", "media/e83ebb4ae76354d99cd9/F4.jpg", "media/e83ebb4ae76354d99cd9/F5.jpg", "media/e83ebb4ae76354d99cd9/F6.jpg", "media/e83ebb4ae76354d99cd9/F7.jpg", "media/e83ebb4ae76354d99cd9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "93fd8df73d31bc8e34ac6bdb52cfdf36ae7d1d71294b0a1e86c1abe14ef61687"}
{"qa_id": "VOST__4198_open_container__camera_translation_direction__00", "video_id": "VOST__4198_open_container", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "up", "backward"], "frame_paths": ["media/c424be1f3cc3cfcd02cf/F0.jpg", "media/c424be1f3cc3cfcd02cf/F1.jpg", "media/c424be1f3cc3cfcd02cf/F2.jpg", "media/c424be1f3cc3cfcd02cf/F3.jpg", "media/c424be1f3cc3cfcd02cf/F4.jpg", "media/c424be1f3cc3cfcd02cf/F5.jpg", "media/c424be1f3cc3cfcd02cf/F6.jpg", "media/c424be1f3cc3cfcd02cf/F7.jpg", "media/c424be1f3cc3cfcd02cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88e05f25e04054bc2ac12ec28033e2a4bb8d9eb719436f2ec5bc816dd3433b46"}
{"qa_id": "VOST__4210_cut_zucchini__camera_translation_direction__00", "video_id": "VOST__4210_cut_zucchini", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "left", "right", "down", "backward"], "frame_paths": ["media/d37593e618428f036f02/F0.jpg", "media/d37593e618428f036f02/F1.jpg", "media/d37593e618428f036f02/F2.jpg", "media/d37593e618428f036f02/F3.jpg", "media/d37593e618428f036f02/F4.jpg", "media/d37593e618428f036f02/F5.jpg", "media/d37593e618428f036f02/F6.jpg", "media/d37593e618428f036f02/F7.jpg", "media/d37593e618428f036f02/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. left\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13e7725622758ded574891de75f3bce76ee8f45675d73dbdca14b7b78db389f5"}
{"qa_id": "VOST__4233_cut_potato__camera_translation_direction__00", "video_id": "VOST__4233_cut_potato", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "left", "right", "up"], "frame_paths": ["media/ca7aff8f57658420588c/F0.jpg", "media/ca7aff8f57658420588c/F1.jpg", "media/ca7aff8f57658420588c/F2.jpg", "media/ca7aff8f57658420588c/F3.jpg", "media/ca7aff8f57658420588c/F4.jpg", "media/ca7aff8f57658420588c/F5.jpg", "media/ca7aff8f57658420588c/F6.jpg", "media/ca7aff8f57658420588c/F7.jpg", "media/ca7aff8f57658420588c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. left\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "85c253b21d8d0c926e696d66ca3a7dc7206d5dbbd05d75301f3541ecda700fe9"}
{"qa_id": "VOST__4233_cut_potato__object_world_motion_direction__01", "video_id": "VOST__4233_cut_potato", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in black shirt with green contour) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "backward", "right", "down"], "frame_paths": ["media/5204732f463915c424e4/F0.jpg", "media/5204732f463915c424e4/F1.jpg", "media/5204732f463915c424e4/F2.jpg", "media/5204732f463915c424e4/F3.jpg", "media/5204732f463915c424e4/F4.jpg", "media/5204732f463915c424e4/F5.jpg", "media/5204732f463915c424e4/F6.jpg", "media/5204732f463915c424e4/F7.jpg", "media/5204732f463915c424e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in black shirt with green contour) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0bd4425cbe18ef2ac7bfe348d65dcd8fd919b1202bd7f1585d046aee1a3ec73b"}
{"qa_id": "VOST__455_fold_box__camera_translation_direction__00", "video_id": "VOST__455_fold_box", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "down", "backward", "right"], "frame_paths": ["media/ea7755c87b75ebb95a5a/F0.jpg", "media/ea7755c87b75ebb95a5a/F1.jpg", "media/ea7755c87b75ebb95a5a/F2.jpg", "media/ea7755c87b75ebb95a5a/F3.jpg", "media/ea7755c87b75ebb95a5a/F4.jpg", "media/ea7755c87b75ebb95a5a/F5.jpg", "media/ea7755c87b75ebb95a5a/F6.jpg", "media/ea7755c87b75ebb95a5a/F7.jpg", "media/ea7755c87b75ebb95a5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1be6f15ccda333035477ae64be76c6ac1279fcd58b6253cd725ff6b2ea438c34"}
{"qa_id": "VOST__4626_cut_paper__camera_translation_direction__00", "video_id": "VOST__4626_cut_paper", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/8ce23f359bd41bfccceb/F0.jpg", "media/8ce23f359bd41bfccceb/F1.jpg", "media/8ce23f359bd41bfccceb/F2.jpg", "media/8ce23f359bd41bfccceb/F3.jpg", "media/8ce23f359bd41bfccceb/F4.jpg", "media/8ce23f359bd41bfccceb/F5.jpg", "media/8ce23f359bd41bfccceb/F6.jpg", "media/8ce23f359bd41bfccceb/F7.jpg", "media/8ce23f359bd41bfccceb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a8eebab78bdd4a112ae9a27b4702791293088707a2542408b18f8f7d65c30b6"}
{"qa_id": "VOST__4632_cut_paper__camera_translation_direction__00", "video_id": "VOST__4632_cut_paper", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "left", "forward", "backward"], "frame_paths": ["media/52e566bd8f59affb79ea/F0.jpg", "media/52e566bd8f59affb79ea/F1.jpg", "media/52e566bd8f59affb79ea/F2.jpg", "media/52e566bd8f59affb79ea/F3.jpg", "media/52e566bd8f59affb79ea/F4.jpg", "media/52e566bd8f59affb79ea/F5.jpg", "media/52e566bd8f59affb79ea/F6.jpg", "media/52e566bd8f59affb79ea/F7.jpg", "media/52e566bd8f59affb79ea/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4b3e78b9cd89381ef642d663d42e99baf75475d5479433368dd5c90325eaec14"}
{"qa_id": "VOST__4632_cut_paper__camera_object_distance_change__01", "video_id": "VOST__4632_cut_paper", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object B (left arm and hand holding a black wire) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/aaddf77d84884a9a92bd/F0.jpg", "media/aaddf77d84884a9a92bd/F1.jpg", "media/aaddf77d84884a9a92bd/F2.jpg", "media/aaddf77d84884a9a92bd/F3.jpg", "media/aaddf77d84884a9a92bd/F4.jpg", "media/aaddf77d84884a9a92bd/F5.jpg", "media/aaddf77d84884a9a92bd/F6.jpg", "media/aaddf77d84884a9a92bd/F7.jpg", "media/aaddf77d84884a9a92bd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (left arm and hand holding a black wire) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1088b3bea50f6f00d5c5ac1c462ccd52e668e0dff44f4a3be2636a6a015edfe3"}
{"qa_id": "VOST__4632_cut_paper__object_pair_distance_change__02", "video_id": "VOST__4632_cut_paper", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did the 3D distance between object A (right arm and hand holding scissors) and object B (left arm and hand holding a black wire) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/98dad63aed6586e391f0/F0.jpg", "media/98dad63aed6586e391f0/F1.jpg", "media/98dad63aed6586e391f0/F2.jpg", "media/98dad63aed6586e391f0/F3.jpg", "media/98dad63aed6586e391f0/F4.jpg", "media/98dad63aed6586e391f0/F5.jpg", "media/98dad63aed6586e391f0/F6.jpg", "media/98dad63aed6586e391f0/F7.jpg", "media/98dad63aed6586e391f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (right arm and hand holding scissors) and object B (left arm and hand holding a black wire) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a1b6c414b16e0666bf01cc171f9343fedaab0028cb82447da3ca6d9891e42b3"}
{"qa_id": "VOST__4663_cut_ginger__camera_translation_direction__00", "video_id": "VOST__4663_cut_ginger", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "left", "forward"], "frame_paths": ["media/13abe82a4b39f54b667e/F0.jpg", "media/13abe82a4b39f54b667e/F1.jpg", "media/13abe82a4b39f54b667e/F2.jpg", "media/13abe82a4b39f54b667e/F3.jpg", "media/13abe82a4b39f54b667e/F4.jpg", "media/13abe82a4b39f54b667e/F5.jpg", "media/13abe82a4b39f54b667e/F6.jpg", "media/13abe82a4b39f54b667e/F7.jpg", "media/13abe82a4b39f54b667e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5055e5ab9f5aeef6ee3cf3b43d2314182c43f6a544a43300b7f0c43dd89748ae"}
{"qa_id": "VOST__4690_cut_guava__camera_translation_direction__00", "video_id": "VOST__4690_cut_guava", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "left", "forward", "backward"], "frame_paths": ["media/1a866125d8d35264f7ee/F0.jpg", "media/1a866125d8d35264f7ee/F1.jpg", "media/1a866125d8d35264f7ee/F2.jpg", "media/1a866125d8d35264f7ee/F3.jpg", "media/1a866125d8d35264f7ee/F4.jpg", "media/1a866125d8d35264f7ee/F5.jpg", "media/1a866125d8d35264f7ee/F6.jpg", "media/1a866125d8d35264f7ee/F7.jpg", "media/1a866125d8d35264f7ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a490d9f3c704dff27e41e6d07a1c4e0065000917882b1cce6eef8035e9a20249"}
{"qa_id": "VOST__4690_cut_guava__camera_view_direction_change__01", "video_id": "VOST__4690_cut_guava", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/1a866125d8d35264f7ee/F0.jpg", "media/1a866125d8d35264f7ee/F1.jpg", "media/1a866125d8d35264f7ee/F2.jpg", "media/1a866125d8d35264f7ee/F3.jpg", "media/1a866125d8d35264f7ee/F4.jpg", "media/1a866125d8d35264f7ee/F5.jpg", "media/1a866125d8d35264f7ee/F6.jpg", "media/1a866125d8d35264f7ee/F7.jpg", "media/1a866125d8d35264f7ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0395f32f9b844f8600c1e21bef7bc7a474df33e6384f9e952c050a4adc2fa1d0"}
{"qa_id": "VOST__4691_cut_guava__camera_translation_direction__00", "video_id": "VOST__4691_cut_guava", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "forward", "backward", "up"], "frame_paths": ["media/6963c557087724506e2e/F0.jpg", "media/6963c557087724506e2e/F1.jpg", "media/6963c557087724506e2e/F2.jpg", "media/6963c557087724506e2e/F3.jpg", "media/6963c557087724506e2e/F4.jpg", "media/6963c557087724506e2e/F5.jpg", "media/6963c557087724506e2e/F6.jpg", "media/6963c557087724506e2e/F7.jpg", "media/6963c557087724506e2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f4247f10c29eaf8ef27d4a935dbcca4fd72efbc58dfa8e0005423bb4ddc7da3d"}
{"qa_id": "VOST__4836_peel_root__camera_translation_direction__00", "video_id": "VOST__4836_peel_root", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "forward", "down"], "frame_paths": ["media/24ac505798238ac17f69/F0.jpg", "media/24ac505798238ac17f69/F1.jpg", "media/24ac505798238ac17f69/F2.jpg", "media/24ac505798238ac17f69/F3.jpg", "media/24ac505798238ac17f69/F4.jpg", "media/24ac505798238ac17f69/F5.jpg", "media/24ac505798238ac17f69/F6.jpg", "media/24ac505798238ac17f69/F7.jpg", "media/24ac505798238ac17f69/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1b63da4199e1166ba6accb610f6260be291628804b23de6379193d9210b3408"}
{"qa_id": "VOST__4900_cut_pumpkin__camera_translation_direction__00", "video_id": "VOST__4900_cut_pumpkin", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "right", "down", "left"], "frame_paths": ["media/f3376d0863135cabf8d1/F0.jpg", "media/f3376d0863135cabf8d1/F1.jpg", "media/f3376d0863135cabf8d1/F2.jpg", "media/f3376d0863135cabf8d1/F3.jpg", "media/f3376d0863135cabf8d1/F4.jpg", "media/f3376d0863135cabf8d1/F5.jpg", "media/f3376d0863135cabf8d1/F6.jpg", "media/f3376d0863135cabf8d1/F7.jpg", "media/f3376d0863135cabf8d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bcde9014174fc84d61ff3ff78dcb035f87c3f824405a09aac652ea94e5a8f570"}
{"qa_id": "VOST__4900_cut_pumpkin__object_travel_distance_comparison__01", "video_id": "VOST__4900_cut_pumpkin", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a light-colored long-sleeve shirt and a black apron) or B (yellow vegetable pieces on a white cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/0c7953b3794dbf8e3329/F0.jpg", "media/0c7953b3794dbf8e3329/F1.jpg", "media/0c7953b3794dbf8e3329/F2.jpg", "media/0c7953b3794dbf8e3329/F3.jpg", "media/0c7953b3794dbf8e3329/F4.jpg", "media/0c7953b3794dbf8e3329/F5.jpg", "media/0c7953b3794dbf8e3329/F6.jpg", "media/0c7953b3794dbf8e3329/F7.jpg", "media/0c7953b3794dbf8e3329/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a light-colored long-sleeve shirt and a black apron) or B (yellow vegetable pieces on a white cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "39641366818136945ebdcb7beff17e21357e7605a80f1a07181d0decf4819a58"}
{"qa_id": "VOST__4981_saw_wood__camera_translation_direction__00", "video_id": "VOST__4981_saw_wood", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "up", "backward", "down", "left"], "frame_paths": ["media/87dcec7df59ad5013a17/F0.jpg", "media/87dcec7df59ad5013a17/F1.jpg", "media/87dcec7df59ad5013a17/F2.jpg", "media/87dcec7df59ad5013a17/F3.jpg", "media/87dcec7df59ad5013a17/F4.jpg", "media/87dcec7df59ad5013a17/F5.jpg", "media/87dcec7df59ad5013a17/F6.jpg", "media/87dcec7df59ad5013a17/F7.jpg", "media/87dcec7df59ad5013a17/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. up\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "922e42f1e49d8ce5c976434417a3f0daaf361acf5d5b9ee03628bcf656485553"}
{"qa_id": "VOST__4981_saw_wood__camera_view_direction_change__01", "video_id": "VOST__4981_saw_wood", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/87dcec7df59ad5013a17/F0.jpg", "media/87dcec7df59ad5013a17/F1.jpg", "media/87dcec7df59ad5013a17/F2.jpg", "media/87dcec7df59ad5013a17/F3.jpg", "media/87dcec7df59ad5013a17/F4.jpg", "media/87dcec7df59ad5013a17/F5.jpg", "media/87dcec7df59ad5013a17/F6.jpg", "media/87dcec7df59ad5013a17/F7.jpg", "media/87dcec7df59ad5013a17/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c05dce671a50983060e3cd70d660ad77033909e622fb6ed995705010d976df5d"}
{"qa_id": "VOST__502_cut_aubergine__camera_translation_direction__00", "video_id": "VOST__502_cut_aubergine", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "up", "forward", "backward"], "frame_paths": ["media/3670a29ac86048044c06/F0.jpg", "media/3670a29ac86048044c06/F1.jpg", "media/3670a29ac86048044c06/F2.jpg", "media/3670a29ac86048044c06/F3.jpg", "media/3670a29ac86048044c06/F4.jpg", "media/3670a29ac86048044c06/F5.jpg", "media/3670a29ac86048044c06/F6.jpg", "media/3670a29ac86048044c06/F7.jpg", "media/3670a29ac86048044c06/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e64b9966eb34732a755ccbfc14e50266f506f664b2b407b9c8f333a7069442bb"}
{"qa_id": "VOST__5033_cut_paper__camera_translation_direction__00", "video_id": "VOST__5033_cut_paper", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "down", "left", "forward", "up"], "frame_paths": ["media/41b365301bee478b13b0/F0.jpg", "media/41b365301bee478b13b0/F1.jpg", "media/41b365301bee478b13b0/F2.jpg", "media/41b365301bee478b13b0/F3.jpg", "media/41b365301bee478b13b0/F4.jpg", "media/41b365301bee478b13b0/F5.jpg", "media/41b365301bee478b13b0/F6.jpg", "media/41b365301bee478b13b0/F7.jpg", "media/41b365301bee478b13b0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. down\nD. left\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "521e608136285bc0bc3a9aee658c4d8dfecb664ff5afbb427e3fa3d933edee9b"}
{"qa_id": "VOST__5074_open_phone__camera_translation_direction__00", "video_id": "VOST__5074_open_phone", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "left", "right", "down"], "frame_paths": ["media/fa97a0ac620017226518/F0.jpg", "media/fa97a0ac620017226518/F1.jpg", "media/fa97a0ac620017226518/F2.jpg", "media/fa97a0ac620017226518/F3.jpg", "media/fa97a0ac620017226518/F4.jpg", "media/fa97a0ac620017226518/F5.jpg", "media/fa97a0ac620017226518/F6.jpg", "media/fa97a0ac620017226518/F7.jpg", "media/fa97a0ac620017226518/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6ef08b51da04b49d5bd7e03b48c51de2309327059aaf82282068da2745bd8cbe"}
{"qa_id": "VOST__5074_open_phone__camera_view_direction_change__01", "video_id": "VOST__5074_open_phone", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/fa97a0ac620017226518/F0.jpg", "media/fa97a0ac620017226518/F1.jpg", "media/fa97a0ac620017226518/F2.jpg", "media/fa97a0ac620017226518/F3.jpg", "media/fa97a0ac620017226518/F4.jpg", "media/fa97a0ac620017226518/F5.jpg", "media/fa97a0ac620017226518/F6.jpg", "media/fa97a0ac620017226518/F7.jpg", "media/fa97a0ac620017226518/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f6cccc833d039220cccadd8b50d4472a612ba19f9450e785b92355da838206d"}
{"qa_id": "VOST__5179_unpack_meat__camera_translation_direction__00", "video_id": "VOST__5179_unpack_meat", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "up", "forward", "right"], "frame_paths": ["media/d865941a46a07f6e23cf/F0.jpg", "media/d865941a46a07f6e23cf/F1.jpg", "media/d865941a46a07f6e23cf/F2.jpg", "media/d865941a46a07f6e23cf/F3.jpg", "media/d865941a46a07f6e23cf/F4.jpg", "media/d865941a46a07f6e23cf/F5.jpg", "media/d865941a46a07f6e23cf/F6.jpg", "media/d865941a46a07f6e23cf/F7.jpg", "media/d865941a46a07f6e23cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "98152f297607ed6643d89c099bdb4bb6572b3d85481e9ff79b1897401cbe190e"}
{"qa_id": "VOST__5179_unpack_meat__object_world_motion_direction__01", "video_id": "VOST__5179_unpack_meat", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person wearing a light purple t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "right", "left", "down", "up"], "frame_paths": ["media/6ad66b2cdf4e8a44b162/F0.jpg", "media/6ad66b2cdf4e8a44b162/F1.jpg", "media/6ad66b2cdf4e8a44b162/F2.jpg", "media/6ad66b2cdf4e8a44b162/F3.jpg", "media/6ad66b2cdf4e8a44b162/F4.jpg", "media/6ad66b2cdf4e8a44b162/F5.jpg", "media/6ad66b2cdf4e8a44b162/F6.jpg", "media/6ad66b2cdf4e8a44b162/F7.jpg", "media/6ad66b2cdf4e8a44b162/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person wearing a light purple t-shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. right\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd82d181177618a503ae19d6c4859c52f35c51a0003bad19e16934a8b61aa368"}
{"qa_id": "VOST__5212_open_blender__camera_translation_direction__00", "video_id": "VOST__5212_open_blender", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/6b9ee8da5bb3f4cc90d8/F0.jpg", "media/6b9ee8da5bb3f4cc90d8/F1.jpg", "media/6b9ee8da5bb3f4cc90d8/F2.jpg", "media/6b9ee8da5bb3f4cc90d8/F3.jpg", "media/6b9ee8da5bb3f4cc90d8/F4.jpg", "media/6b9ee8da5bb3f4cc90d8/F5.jpg", "media/6b9ee8da5bb3f4cc90d8/F6.jpg", "media/6b9ee8da5bb3f4cc90d8/F7.jpg", "media/6b9ee8da5bb3f4cc90d8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fbaa2d78eef1ebc22787e2728f3a6960dae64b5fe959bcc2ca0a2b2b423c8fc8"}
{"qa_id": "VOST__5253_open_package__camera_translation_direction__00", "video_id": "VOST__5253_open_package", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "right", "left", "up"], "frame_paths": ["media/4210f28c725627c19c8d/F0.jpg", "media/4210f28c725627c19c8d/F1.jpg", "media/4210f28c725627c19c8d/F2.jpg", "media/4210f28c725627c19c8d/F3.jpg", "media/4210f28c725627c19c8d/F4.jpg", "media/4210f28c725627c19c8d/F5.jpg", "media/4210f28c725627c19c8d/F6.jpg", "media/4210f28c725627c19c8d/F7.jpg", "media/4210f28c725627c19c8d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b8942d980c77b4b4cf66881523e5714f0a7fdf4af07a7ab52a40b16beb2e4c7"}
{"qa_id": "VOST__5263_open_engine__camera_translation_direction__00", "video_id": "VOST__5263_open_engine", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "forward", "right", "up"], "frame_paths": ["media/8e36e6bf30ee42cdc13c/F0.jpg", "media/8e36e6bf30ee42cdc13c/F1.jpg", "media/8e36e6bf30ee42cdc13c/F2.jpg", "media/8e36e6bf30ee42cdc13c/F3.jpg", "media/8e36e6bf30ee42cdc13c/F4.jpg", "media/8e36e6bf30ee42cdc13c/F5.jpg", "media/8e36e6bf30ee42cdc13c/F6.jpg", "media/8e36e6bf30ee42cdc13c/F7.jpg", "media/8e36e6bf30ee42cdc13c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. forward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "169d4f7b4d4eb8c21028175adc0e5737bb1195fdcd9439f8ae6179681b48d3dd"}
{"qa_id": "VOST__5263_open_engine__camera_view_direction_change__01", "video_id": "VOST__5263_open_engine", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/8e36e6bf30ee42cdc13c/F0.jpg", "media/8e36e6bf30ee42cdc13c/F1.jpg", "media/8e36e6bf30ee42cdc13c/F2.jpg", "media/8e36e6bf30ee42cdc13c/F3.jpg", "media/8e36e6bf30ee42cdc13c/F4.jpg", "media/8e36e6bf30ee42cdc13c/F5.jpg", "media/8e36e6bf30ee42cdc13c/F6.jpg", "media/8e36e6bf30ee42cdc13c/F7.jpg", "media/8e36e6bf30ee42cdc13c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1c59a0544288e7552ea4a6206c32e2f7efbf8148159ac90187434562a607bd02"}
{"qa_id": "VOST__5288_open_box__camera_translation_direction__00", "video_id": "VOST__5288_open_box", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "up", "right", "forward", "left"], "frame_paths": ["media/764404e6a5f213339089/F0.jpg", "media/764404e6a5f213339089/F1.jpg", "media/764404e6a5f213339089/F2.jpg", "media/764404e6a5f213339089/F3.jpg", "media/764404e6a5f213339089/F4.jpg", "media/764404e6a5f213339089/F5.jpg", "media/764404e6a5f213339089/F6.jpg", "media/764404e6a5f213339089/F7.jpg", "media/764404e6a5f213339089/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. up\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4754a58e0376444151a88b0d55811d6afa4064e8cfce479079c8861e640b0535"}
{"qa_id": "VOST__531_cut_potato__camera_translation_direction__00", "video_id": "VOST__531_cut_potato", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "up", "forward", "right"], "frame_paths": ["media/0e43959230aa6a9d50ea/F0.jpg", "media/0e43959230aa6a9d50ea/F1.jpg", "media/0e43959230aa6a9d50ea/F2.jpg", "media/0e43959230aa6a9d50ea/F3.jpg", "media/0e43959230aa6a9d50ea/F4.jpg", "media/0e43959230aa6a9d50ea/F5.jpg", "media/0e43959230aa6a9d50ea/F6.jpg", "media/0e43959230aa6a9d50ea/F7.jpg", "media/0e43959230aa6a9d50ea/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "940713b2439bed7b057e6db20045cd67a8705dfeab0f3143614e7cb1501bf8c5"}
{"qa_id": "VOST__5323_open_container__camera_translation_direction__00", "video_id": "VOST__5323_open_container", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "left", "right", "up"], "frame_paths": ["media/44527e6453f6a92e9391/F0.jpg", "media/44527e6453f6a92e9391/F1.jpg", "media/44527e6453f6a92e9391/F2.jpg", "media/44527e6453f6a92e9391/F3.jpg", "media/44527e6453f6a92e9391/F4.jpg", "media/44527e6453f6a92e9391/F5.jpg", "media/44527e6453f6a92e9391/F6.jpg", "media/44527e6453f6a92e9391/F7.jpg", "media/44527e6453f6a92e9391/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. left\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "12c913f04103660d8410f2bfc58b6be5a249749b4e2c40f37bb7cffd9d08c8ed"}
{"qa_id": "VOST__532_cut_potato__camera_translation_direction__00", "video_id": "VOST__532_cut_potato", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "up", "left", "backward", "right"], "frame_paths": ["media/b0d1ca0376ef3a2922db/F0.jpg", "media/b0d1ca0376ef3a2922db/F1.jpg", "media/b0d1ca0376ef3a2922db/F2.jpg", "media/b0d1ca0376ef3a2922db/F3.jpg", "media/b0d1ca0376ef3a2922db/F4.jpg", "media/b0d1ca0376ef3a2922db/F5.jpg", "media/b0d1ca0376ef3a2922db/F6.jpg", "media/b0d1ca0376ef3a2922db/F7.jpg", "media/b0d1ca0376ef3a2922db/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. up\nD. left\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f2563240521bbecbec70388e89aed33e9b8952abe66d0d8e8ac5e075dc4e3a15"}
{"qa_id": "VOST__533_cut_asparagus__camera_translation_direction__00", "video_id": "VOST__533_cut_asparagus", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "left", "down", "right", "forward"], "frame_paths": ["media/c97c888cf03d3e9039d7/F0.jpg", "media/c97c888cf03d3e9039d7/F1.jpg", "media/c97c888cf03d3e9039d7/F2.jpg", "media/c97c888cf03d3e9039d7/F3.jpg", "media/c97c888cf03d3e9039d7/F4.jpg", "media/c97c888cf03d3e9039d7/F5.jpg", "media/c97c888cf03d3e9039d7/F6.jpg", "media/c97c888cf03d3e9039d7/F7.jpg", "media/c97c888cf03d3e9039d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. left\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48d932b81549d68d81adde43823a9a61f519a0e8c0287680c3cd542bcaa58a9b"}
{"qa_id": "VOST__533_cut_asparagus__camera_view_direction_change__01", "video_id": "VOST__533_cut_asparagus", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/c97c888cf03d3e9039d7/F0.jpg", "media/c97c888cf03d3e9039d7/F1.jpg", "media/c97c888cf03d3e9039d7/F2.jpg", "media/c97c888cf03d3e9039d7/F3.jpg", "media/c97c888cf03d3e9039d7/F4.jpg", "media/c97c888cf03d3e9039d7/F5.jpg", "media/c97c888cf03d3e9039d7/F6.jpg", "media/c97c888cf03d3e9039d7/F7.jpg", "media/c97c888cf03d3e9039d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb821ef5ac91183e7c2873f4974564eb33a0896e76a359527a26534ed1c81dfe"}
{"qa_id": "VOST__5342_open_bag__camera_translation_direction__00", "video_id": "VOST__5342_open_bag", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "left", "right", "up", "forward"], "frame_paths": ["media/27f654c93d60474177a6/F0.jpg", "media/27f654c93d60474177a6/F1.jpg", "media/27f654c93d60474177a6/F2.jpg", "media/27f654c93d60474177a6/F3.jpg", "media/27f654c93d60474177a6/F4.jpg", "media/27f654c93d60474177a6/F5.jpg", "media/27f654c93d60474177a6/F6.jpg", "media/27f654c93d60474177a6/F7.jpg", "media/27f654c93d60474177a6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. left\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab0dac4300b3a2f663f104863fad79da43ab5931f671f87a8278021176e693dd"}
{"qa_id": "VOST__5783_apply_paint__camera_translation_direction__00", "video_id": "VOST__5783_apply_paint", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "up", "forward", "right"], "frame_paths": ["media/b79429b4beb8cae58cb0/F0.jpg", "media/b79429b4beb8cae58cb0/F1.jpg", "media/b79429b4beb8cae58cb0/F2.jpg", "media/b79429b4beb8cae58cb0/F3.jpg", "media/b79429b4beb8cae58cb0/F4.jpg", "media/b79429b4beb8cae58cb0/F5.jpg", "media/b79429b4beb8cae58cb0/F6.jpg", "media/b79429b4beb8cae58cb0/F7.jpg", "media/b79429b4beb8cae58cb0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "92cd5bf3ebd2a353124ba769bbd30b9fefbb66e9ba995a8886d2d1cd51455573"}
{"qa_id": "VOST__5788_apply_paint__camera_translation_direction__00", "video_id": "VOST__5788_apply_paint", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/b21f0d0c365dd21dc4a8/F0.jpg", "media/b21f0d0c365dd21dc4a8/F1.jpg", "media/b21f0d0c365dd21dc4a8/F2.jpg", "media/b21f0d0c365dd21dc4a8/F3.jpg", "media/b21f0d0c365dd21dc4a8/F4.jpg", "media/b21f0d0c365dd21dc4a8/F5.jpg", "media/b21f0d0c365dd21dc4a8/F6.jpg", "media/b21f0d0c365dd21dc4a8/F7.jpg", "media/b21f0d0c365dd21dc4a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5a7600045a78368e27c66685b2f35d6e0a8a381470bbe1063cbdb206f47c831"}
{"qa_id": "VOST__6077_wash_cloth__camera_translation_direction__00", "video_id": "VOST__6077_wash_cloth", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "down", "up", "left"], "frame_paths": ["media/733086f465a37fa2b004/F0.jpg", "media/733086f465a37fa2b004/F1.jpg", "media/733086f465a37fa2b004/F2.jpg", "media/733086f465a37fa2b004/F3.jpg", "media/733086f465a37fa2b004/F4.jpg", "media/733086f465a37fa2b004/F5.jpg", "media/733086f465a37fa2b004/F6.jpg", "media/733086f465a37fa2b004/F7.jpg", "media/733086f465a37fa2b004/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efe4e33866650de14aa38cf6e9590f462bab45fca0ffcb1a44ec1929d299194c"}
{"qa_id": "VOST__6099_spread_glue__camera_translation_direction__00", "video_id": "VOST__6099_spread_glue", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "left", "right", "backward"], "frame_paths": ["media/11d14804a54595734587/F0.jpg", "media/11d14804a54595734587/F1.jpg", "media/11d14804a54595734587/F2.jpg", "media/11d14804a54595734587/F3.jpg", "media/11d14804a54595734587/F4.jpg", "media/11d14804a54595734587/F5.jpg", "media/11d14804a54595734587/F6.jpg", "media/11d14804a54595734587/F7.jpg", "media/11d14804a54595734587/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66755c8ab2fb97b716a2e88bbb1afddefc7dce1599b2a87cabbf531ef22a6cc5"}
{"qa_id": "VOST__6099_spread_glue__object_world_motion_direction__01", "video_id": "VOST__6099_spread_glue", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person's right arm and hand holding a tool) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "up", "forward", "down"], "frame_paths": ["media/3df2c02db1cc52ec9372/F0.jpg", "media/3df2c02db1cc52ec9372/F1.jpg", "media/3df2c02db1cc52ec9372/F2.jpg", "media/3df2c02db1cc52ec9372/F3.jpg", "media/3df2c02db1cc52ec9372/F4.jpg", "media/3df2c02db1cc52ec9372/F5.jpg", "media/3df2c02db1cc52ec9372/F6.jpg", "media/3df2c02db1cc52ec9372/F7.jpg", "media/3df2c02db1cc52ec9372/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person's right arm and hand holding a tool) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82ad34f958cd4ce53513a482ea8fde8affc98a92d89cf518d827f9892e268a31"}
{"qa_id": "VOST__6099_spread_glue__object_travel_distance_comparison__02", "video_id": "VOST__6099_spread_glue", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a tool) or B (person's left arm and hand pressing down on cardboard)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/7a78e549031f8c1f4da8/F0.jpg", "media/7a78e549031f8c1f4da8/F1.jpg", "media/7a78e549031f8c1f4da8/F2.jpg", "media/7a78e549031f8c1f4da8/F3.jpg", "media/7a78e549031f8c1f4da8/F4.jpg", "media/7a78e549031f8c1f4da8/F5.jpg", "media/7a78e549031f8c1f4da8/F6.jpg", "media/7a78e549031f8c1f4da8/F7.jpg", "media/7a78e549031f8c1f4da8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a tool) or B (person's left arm and hand pressing down on cardboard)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654c0032017fa7041241a7bb63b476deb6ddb449fa888cc84895177be8274e2a"}
{"qa_id": "VOST__6101_spread_glue__camera_translation_direction__00", "video_id": "VOST__6101_spread_glue", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "up", "right", "forward"], "frame_paths": ["media/1994b2c4df0cb3624543/F0.jpg", "media/1994b2c4df0cb3624543/F1.jpg", "media/1994b2c4df0cb3624543/F2.jpg", "media/1994b2c4df0cb3624543/F3.jpg", "media/1994b2c4df0cb3624543/F4.jpg", "media/1994b2c4df0cb3624543/F5.jpg", "media/1994b2c4df0cb3624543/F6.jpg", "media/1994b2c4df0cb3624543/F7.jpg", "media/1994b2c4df0cb3624543/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3724fe15e10fd9c0950a2295d3284ec375d95c9a81a61ac820b4061d1f369ea3"}
{"qa_id": "VOST__6152_knead_dough__camera_translation_direction__00", "video_id": "VOST__6152_knead_dough", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "forward", "down", "left"], "frame_paths": ["media/6801c9258bd3cee6a00c/F0.jpg", "media/6801c9258bd3cee6a00c/F1.jpg", "media/6801c9258bd3cee6a00c/F2.jpg", "media/6801c9258bd3cee6a00c/F3.jpg", "media/6801c9258bd3cee6a00c/F4.jpg", "media/6801c9258bd3cee6a00c/F5.jpg", "media/6801c9258bd3cee6a00c/F6.jpg", "media/6801c9258bd3cee6a00c/F7.jpg", "media/6801c9258bd3cee6a00c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d555a863c376ef25307b6a5fce72a0735f5a0be8d7669e6f171b89d3b5e5ea15"}
{"qa_id": "VOST__6209_apply_paint__camera_translation_direction__00", "video_id": "VOST__6209_apply_paint", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "left", "backward", "down", "up"], "frame_paths": ["media/9da7cdea8f10e212f76b/F0.jpg", "media/9da7cdea8f10e212f76b/F1.jpg", "media/9da7cdea8f10e212f76b/F2.jpg", "media/9da7cdea8f10e212f76b/F3.jpg", "media/9da7cdea8f10e212f76b/F4.jpg", "media/9da7cdea8f10e212f76b/F5.jpg", "media/9da7cdea8f10e212f76b/F6.jpg", "media/9da7cdea8f10e212f76b/F7.jpg", "media/9da7cdea8f10e212f76b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. left\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5e6829967cc08a74dcda93d75202287effeef930d209ecdb0e6f7cca58c7e2b"}
{"qa_id": "VOST__6268_apply_paint__camera_translation_direction__00", "video_id": "VOST__6268_apply_paint", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "down", "right"], "frame_paths": ["media/8827dd557aa96e8fa63d/F0.jpg", "media/8827dd557aa96e8fa63d/F1.jpg", "media/8827dd557aa96e8fa63d/F2.jpg", "media/8827dd557aa96e8fa63d/F3.jpg", "media/8827dd557aa96e8fa63d/F4.jpg", "media/8827dd557aa96e8fa63d/F5.jpg", "media/8827dd557aa96e8fa63d/F6.jpg", "media/8827dd557aa96e8fa63d/F7.jpg", "media/8827dd557aa96e8fa63d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f6a1d99413469b6dcea520cc29857359f8a98c3e6d135cfd5f6f7ed5d46f07df"}
{"qa_id": "VOST__6268_apply_paint__object_world_motion_direction__01", "video_id": "VOST__6268_apply_paint", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (hand holding a paintbrush) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "up", "forward", "right"], "frame_paths": ["media/6aa31737a108e1526985/F0.jpg", "media/6aa31737a108e1526985/F1.jpg", "media/6aa31737a108e1526985/F2.jpg", "media/6aa31737a108e1526985/F3.jpg", "media/6aa31737a108e1526985/F4.jpg", "media/6aa31737a108e1526985/F5.jpg", "media/6aa31737a108e1526985/F6.jpg", "media/6aa31737a108e1526985/F7.jpg", "media/6aa31737a108e1526985/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (hand holding a paintbrush) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7b84eb0a5c440188e26baed53ef9d245d54d03198c22251b6c6bd72d70981d2d"}
{"qa_id": "VOST__6268_apply_paint__object_pair_distance_change__02", "video_id": "VOST__6268_apply_paint", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did the 3D distance between object A (hand holding a paintbrush) and object B (person wearing a plaid shirt and a smartwatch) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/0a6468599573d68b6102/F0.jpg", "media/0a6468599573d68b6102/F1.jpg", "media/0a6468599573d68b6102/F2.jpg", "media/0a6468599573d68b6102/F3.jpg", "media/0a6468599573d68b6102/F4.jpg", "media/0a6468599573d68b6102/F5.jpg", "media/0a6468599573d68b6102/F6.jpg", "media/0a6468599573d68b6102/F7.jpg", "media/0a6468599573d68b6102/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (hand holding a paintbrush) and object B (person wearing a plaid shirt and a smartwatch) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "32aadc8205dc3bb0478981f2f3da71ceab0d772aa9e089eba637c29dea468306"}
{"qa_id": "VOST__6315_apply_paint__object_world_motion_direction__00", "video_id": "VOST__6315_apply_paint", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (person's arm wearing a black glove and orange sleeve) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "down", "left", "up"], "frame_paths": ["media/0be52baa9c7160d12460/F0.jpg", "media/0be52baa9c7160d12460/F1.jpg", "media/0be52baa9c7160d12460/F2.jpg", "media/0be52baa9c7160d12460/F3.jpg", "media/0be52baa9c7160d12460/F4.jpg", "media/0be52baa9c7160d12460/F5.jpg", "media/0be52baa9c7160d12460/F6.jpg", "media/0be52baa9c7160d12460/F7.jpg", "media/0be52baa9c7160d12460/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (person's arm wearing a black glove and orange sleeve) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13359330738e76ac745a9100bb3b32a10401466390f4b782059fbb47c850f2d0"}
{"qa_id": "VOST__6315_apply_paint__object_pair_distance_change__01", "video_id": "VOST__6315_apply_paint", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did the 3D distance between object A (person's arm wearing a black glove and orange sleeve) and object B (pair of grey sneakers with white laces) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8528cd6bbaf33958c707/F0.jpg", "media/8528cd6bbaf33958c707/F1.jpg", "media/8528cd6bbaf33958c707/F2.jpg", "media/8528cd6bbaf33958c707/F3.jpg", "media/8528cd6bbaf33958c707/F4.jpg", "media/8528cd6bbaf33958c707/F5.jpg", "media/8528cd6bbaf33958c707/F6.jpg", "media/8528cd6bbaf33958c707/F7.jpg", "media/8528cd6bbaf33958c707/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (person's arm wearing a black glove and orange sleeve) and object B (pair of grey sneakers with white laces) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a02fe6d4870e566237555ccc5a2a0c0a00e32f863865ce6d474c807f7fcd6b08"}
{"qa_id": "VOST__6315_apply_paint__object_travel_distance_comparison__02", "video_id": "VOST__6315_apply_paint", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's arm wearing a black glove and orange sleeve) or B (pair of grey sneakers with white laces)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/8528cd6bbaf33958c707/F0.jpg", "media/8528cd6bbaf33958c707/F1.jpg", "media/8528cd6bbaf33958c707/F2.jpg", "media/8528cd6bbaf33958c707/F3.jpg", "media/8528cd6bbaf33958c707/F4.jpg", "media/8528cd6bbaf33958c707/F5.jpg", "media/8528cd6bbaf33958c707/F6.jpg", "media/8528cd6bbaf33958c707/F7.jpg", "media/8528cd6bbaf33958c707/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's arm wearing a black glove and orange sleeve) or B (pair of grey sneakers with white laces)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2ff030c7a69ad974fafea9f84676842b7a4bd4680eeaccf64180ee16ec469d3b"}
{"qa_id": "VOST__6342_apply_paint__camera_translation_direction__00", "video_id": "VOST__6342_apply_paint", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "down", "left", "up"], "frame_paths": ["media/026c072ff0b801c89cc6/F0.jpg", "media/026c072ff0b801c89cc6/F1.jpg", "media/026c072ff0b801c89cc6/F2.jpg", "media/026c072ff0b801c89cc6/F3.jpg", "media/026c072ff0b801c89cc6/F4.jpg", "media/026c072ff0b801c89cc6/F5.jpg", "media/026c072ff0b801c89cc6/F6.jpg", "media/026c072ff0b801c89cc6/F7.jpg", "media/026c072ff0b801c89cc6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3283ce3d7a0fe4bd0e99c8ad13a616b184db01f8bd534889843ec8b6b5e9415"}
{"qa_id": "VOST__6342_apply_paint__camera_object_distance_change__01", "video_id": "VOST__6342_apply_paint", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, did object B (small grey miniature figure) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c461fd97c77e9ddc2063/F0.jpg", "media/c461fd97c77e9ddc2063/F1.jpg", "media/c461fd97c77e9ddc2063/F2.jpg", "media/c461fd97c77e9ddc2063/F3.jpg", "media/c461fd97c77e9ddc2063/F4.jpg", "media/c461fd97c77e9ddc2063/F5.jpg", "media/c461fd97c77e9ddc2063/F6.jpg", "media/c461fd97c77e9ddc2063/F7.jpg", "media/c461fd97c77e9ddc2063/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object B (small grey miniature figure) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "61ee89ecbc6660a5fda500c1539a048a1e6c65032733894c9e57e7ea7d0e9419"}
{"qa_id": "VOST__6346_apply_paint__camera_translation_direction__00", "video_id": "VOST__6346_apply_paint", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "left", "right", "backward"], "frame_paths": ["media/7594a7c2868e0145a2b1/F0.jpg", "media/7594a7c2868e0145a2b1/F1.jpg", "media/7594a7c2868e0145a2b1/F2.jpg", "media/7594a7c2868e0145a2b1/F3.jpg", "media/7594a7c2868e0145a2b1/F4.jpg", "media/7594a7c2868e0145a2b1/F5.jpg", "media/7594a7c2868e0145a2b1/F6.jpg", "media/7594a7c2868e0145a2b1/F7.jpg", "media/7594a7c2868e0145a2b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9abaddfd31249831c1ecc2c887da7dcf4205d24b66d591f85de426ec88bdd92b"}
{"qa_id": "VOST__6404_apply_pencil__camera_translation_direction__00", "video_id": "VOST__6404_apply_pencil", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "backward", "up"], "frame_paths": ["media/76a11cd352633f1029d6/F0.jpg", "media/76a11cd352633f1029d6/F1.jpg", "media/76a11cd352633f1029d6/F2.jpg", "media/76a11cd352633f1029d6/F3.jpg", "media/76a11cd352633f1029d6/F4.jpg", "media/76a11cd352633f1029d6/F5.jpg", "media/76a11cd352633f1029d6/F6.jpg", "media/76a11cd352633f1029d6/F7.jpg", "media/76a11cd352633f1029d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "53eeed872bcd1b1b0a395669e57e035558e3f4c7bfb297226548babe7a3ff392"}
{"qa_id": "VOST__6404_apply_pencil__object_world_motion_direction__01", "video_id": "VOST__6404_apply_pencil", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (left hand resting on the paper) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "left", "right", "down", "backward"], "frame_paths": ["media/f9c424031a470e9a2dd9/F0.jpg", "media/f9c424031a470e9a2dd9/F1.jpg", "media/f9c424031a470e9a2dd9/F2.jpg", "media/f9c424031a470e9a2dd9/F3.jpg", "media/f9c424031a470e9a2dd9/F4.jpg", "media/f9c424031a470e9a2dd9/F5.jpg", "media/f9c424031a470e9a2dd9/F6.jpg", "media/f9c424031a470e9a2dd9/F7.jpg", "media/f9c424031a470e9a2dd9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (left hand resting on the paper) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. left\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "58a6c1bd3dba3b63eb4930718a40526fcf01ac86b2cb93cf561d5a701d407209"}
{"qa_id": "VOST__6404_apply_pencil__camera_object_distance_change__02", "video_id": "VOST__6404_apply_pencil", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object B (left hand resting on the paper) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/f9c424031a470e9a2dd9/F0.jpg", "media/f9c424031a470e9a2dd9/F1.jpg", "media/f9c424031a470e9a2dd9/F2.jpg", "media/f9c424031a470e9a2dd9/F3.jpg", "media/f9c424031a470e9a2dd9/F4.jpg", "media/f9c424031a470e9a2dd9/F5.jpg", "media/f9c424031a470e9a2dd9/F6.jpg", "media/f9c424031a470e9a2dd9/F7.jpg", "media/f9c424031a470e9a2dd9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (left hand resting on the paper) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a94e6f55c137429f357e8add2dfec85441bbad3b68d6a3366c1fca2735ab9e5"}
{"qa_id": "VOST__6404_apply_pencil__object_pair_distance_change__03", "video_id": "VOST__6404_apply_pencil", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (right hand holding a blue colored pencil) and object B (left hand resting on the paper) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/6c8981d025d41e712da9/F0.jpg", "media/6c8981d025d41e712da9/F1.jpg", "media/6c8981d025d41e712da9/F2.jpg", "media/6c8981d025d41e712da9/F3.jpg", "media/6c8981d025d41e712da9/F4.jpg", "media/6c8981d025d41e712da9/F5.jpg", "media/6c8981d025d41e712da9/F6.jpg", "media/6c8981d025d41e712da9/F7.jpg", "media/6c8981d025d41e712da9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (right hand holding a blue colored pencil) and object B (left hand resting on the paper) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6146c308e1727ea6658c7201c5ef1b85b77a3747df5a71d012964f22a67668c5"}
{"qa_id": "VOST__6407_apply_pencil__object_world_motion_direction__00", "video_id": "VOST__6407_apply_pencil", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (right hand holding a pencil) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "down", "right", "forward"], "frame_paths": ["media/e406ca57821798188e5f/F0.jpg", "media/e406ca57821798188e5f/F1.jpg", "media/e406ca57821798188e5f/F2.jpg", "media/e406ca57821798188e5f/F3.jpg", "media/e406ca57821798188e5f/F4.jpg", "media/e406ca57821798188e5f/F5.jpg", "media/e406ca57821798188e5f/F6.jpg", "media/e406ca57821798188e5f/F7.jpg", "media/e406ca57821798188e5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (right hand holding a pencil) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f12c726100ef65ea4a53b6713785e6aed19a0aab0cd58da3f70f9901318250bb"}
{"qa_id": "VOST__6407_apply_pencil__object_travel_distance_comparison__01", "video_id": "VOST__6407_apply_pencil", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a pencil) or B (left hand wearing a black watch)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/b69c2a17cf25a6d3df49/F0.jpg", "media/b69c2a17cf25a6d3df49/F1.jpg", "media/b69c2a17cf25a6d3df49/F2.jpg", "media/b69c2a17cf25a6d3df49/F3.jpg", "media/b69c2a17cf25a6d3df49/F4.jpg", "media/b69c2a17cf25a6d3df49/F5.jpg", "media/b69c2a17cf25a6d3df49/F6.jpg", "media/b69c2a17cf25a6d3df49/F7.jpg", "media/b69c2a17cf25a6d3df49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a pencil) or B (left hand wearing a black watch)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8fdcf3a3d85bf920eafedd8034723ec0ec872ffe608cb8ac3d868eed506f984a"}
{"qa_id": "VOST__6409_paint_plate__camera_translation_direction__00", "video_id": "VOST__6409_paint_plate", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "up", "forward", "down"], "frame_paths": ["media/449c595e2e07ade3f8e8/F0.jpg", "media/449c595e2e07ade3f8e8/F1.jpg", "media/449c595e2e07ade3f8e8/F2.jpg", "media/449c595e2e07ade3f8e8/F3.jpg", "media/449c595e2e07ade3f8e8/F4.jpg", "media/449c595e2e07ade3f8e8/F5.jpg", "media/449c595e2e07ade3f8e8/F6.jpg", "media/449c595e2e07ade3f8e8/F7.jpg", "media/449c595e2e07ade3f8e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "46ddff8101658d9f6f12a80ba35263e0b576ea122fca95a8e3d56815cbfd308d"}
{"qa_id": "VOST__6417_apply_pencil__camera_translation_direction__00", "video_id": "VOST__6417_apply_pencil", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "backward", "down", "right"], "frame_paths": ["media/b07784640048fbd7841a/F0.jpg", "media/b07784640048fbd7841a/F1.jpg", "media/b07784640048fbd7841a/F2.jpg", "media/b07784640048fbd7841a/F3.jpg", "media/b07784640048fbd7841a/F4.jpg", "media/b07784640048fbd7841a/F5.jpg", "media/b07784640048fbd7841a/F6.jpg", "media/b07784640048fbd7841a/F7.jpg", "media/b07784640048fbd7841a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. backward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1ef5279ae5ed323a527b64a04511fb83fc0a2ae92047513ea70cf92d4bda6ed8"}
{"qa_id": "VOST__6417_apply_pencil__camera_view_direction_change__01", "video_id": "VOST__6417_apply_pencil", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/b07784640048fbd7841a/F0.jpg", "media/b07784640048fbd7841a/F1.jpg", "media/b07784640048fbd7841a/F2.jpg", "media/b07784640048fbd7841a/F3.jpg", "media/b07784640048fbd7841a/F4.jpg", "media/b07784640048fbd7841a/F5.jpg", "media/b07784640048fbd7841a/F6.jpg", "media/b07784640048fbd7841a/F7.jpg", "media/b07784640048fbd7841a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c00a9377eb439677542cfd9d8ba66bdf6ed121853ced4e519c06dbe9ab83159e"}
{"qa_id": "VOST__6474_apply_paint__camera_view_direction_change__00", "video_id": "VOST__6474_apply_paint", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/bfc18e0697007f3d5ac5/F0.jpg", "media/bfc18e0697007f3d5ac5/F1.jpg", "media/bfc18e0697007f3d5ac5/F2.jpg", "media/bfc18e0697007f3d5ac5/F3.jpg", "media/bfc18e0697007f3d5ac5/F4.jpg", "media/bfc18e0697007f3d5ac5/F5.jpg", "media/bfc18e0697007f3d5ac5/F6.jpg", "media/bfc18e0697007f3d5ac5/F7.jpg", "media/bfc18e0697007f3d5ac5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13e8a1acb3bede661435ab91735c5cbbfe935d0cd5c1b011dda4a116b80e108c"}
{"qa_id": "VOST__6489_apply_paint__camera_translation_direction__00", "video_id": "VOST__6489_apply_paint", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "right", "down", "left"], "frame_paths": ["media/99046569d5b53a852991/F0.jpg", "media/99046569d5b53a852991/F1.jpg", "media/99046569d5b53a852991/F2.jpg", "media/99046569d5b53a852991/F3.jpg", "media/99046569d5b53a852991/F4.jpg", "media/99046569d5b53a852991/F5.jpg", "media/99046569d5b53a852991/F6.jpg", "media/99046569d5b53a852991/F7.jpg", "media/99046569d5b53a852991/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0625051e546d3c4806b189e2587be65800f8ad08cc6accf1b00221b36b9c85dd"}
{"qa_id": "VOST__6490_apply_paint__camera_translation_direction__00", "video_id": "VOST__6490_apply_paint", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "down", "left", "right", "backward"], "frame_paths": ["media/5dc165b1373d0005a390/F0.jpg", "media/5dc165b1373d0005a390/F1.jpg", "media/5dc165b1373d0005a390/F2.jpg", "media/5dc165b1373d0005a390/F3.jpg", "media/5dc165b1373d0005a390/F4.jpg", "media/5dc165b1373d0005a390/F5.jpg", "media/5dc165b1373d0005a390/F6.jpg", "media/5dc165b1373d0005a390/F7.jpg", "media/5dc165b1373d0005a390/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. down\nD. left\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bfde84afea319d9c10533bc6f0c6b4292d7465aab18886b23803dbbed60d2e41"}
{"qa_id": "VOST__6863_spread_cards__camera_translation_direction__00", "video_id": "VOST__6863_spread_cards", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "up", "right", "backward"], "frame_paths": ["media/78a588aed3e66f98c8dd/F0.jpg", "media/78a588aed3e66f98c8dd/F1.jpg", "media/78a588aed3e66f98c8dd/F2.jpg", "media/78a588aed3e66f98c8dd/F3.jpg", "media/78a588aed3e66f98c8dd/F4.jpg", "media/78a588aed3e66f98c8dd/F5.jpg", "media/78a588aed3e66f98c8dd/F6.jpg", "media/78a588aed3e66f98c8dd/F7.jpg", "media/78a588aed3e66f98c8dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. up\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ea74062b708c168ff54b989f4737185343eeca4ba2ead68440c8beab2b975cb0"}
{"qa_id": "VOST__700_peel_onion__object_travel_distance_comparison__00", "video_id": "VOST__700_peel_onion", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding an onion on a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/30c6f2e8233eb1e55270/F0.jpg", "media/30c6f2e8233eb1e55270/F1.jpg", "media/30c6f2e8233eb1e55270/F2.jpg", "media/30c6f2e8233eb1e55270/F3.jpg", "media/30c6f2e8233eb1e55270/F4.jpg", "media/30c6f2e8233eb1e55270/F5.jpg", "media/30c6f2e8233eb1e55270/F6.jpg", "media/30c6f2e8233eb1e55270/F7.jpg", "media/30c6f2e8233eb1e55270/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right hand holding a knife) or B (left hand holding an onion on a cutting board)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "599fccc6230a72ef6bd4e8d269638913e17b25a3b9ef88cbfc2625e28eb0f0c2"}
{"qa_id": "VOST__707_cut_butter__camera_translation_direction__00", "video_id": "VOST__707_cut_butter", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "forward", "left"], "frame_paths": ["media/0599568559b5caa1745a/F0.jpg", "media/0599568559b5caa1745a/F1.jpg", "media/0599568559b5caa1745a/F2.jpg", "media/0599568559b5caa1745a/F3.jpg", "media/0599568559b5caa1745a/F4.jpg", "media/0599568559b5caa1745a/F5.jpg", "media/0599568559b5caa1745a/F6.jpg", "media/0599568559b5caa1745a/F7.jpg", "media/0599568559b5caa1745a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c7e8ab4bc37ff0bcd1f52ac3c5cb5ef21f9f5f6367f28061c7de2e1c4c40ab0"}
{"qa_id": "VOST__7088_spread_cards__camera_translation_direction__00", "video_id": "VOST__7088_spread_cards", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "up", "right", "backward"], "frame_paths": ["media/18cea488f77e848ff373/F0.jpg", "media/18cea488f77e848ff373/F1.jpg", "media/18cea488f77e848ff373/F2.jpg", "media/18cea488f77e848ff373/F3.jpg", "media/18cea488f77e848ff373/F4.jpg", "media/18cea488f77e848ff373/F5.jpg", "media/18cea488f77e848ff373/F6.jpg", "media/18cea488f77e848ff373/F7.jpg", "media/18cea488f77e848ff373/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. up\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7eb4a111ceb32ba456933e7513cd755e5d4d39798b41a48df2691e6f6418622"}
{"qa_id": "VOST__7088_spread_cards__object_pair_distance_change__01", "video_id": "VOST__7088_spread_cards", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (right hand of a person wearing an orange long-sleeved shirt) and object B (left hand of a person wearing an orange long-sleeved shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/381feef28fd8be651d7c/F0.jpg", "media/381feef28fd8be651d7c/F1.jpg", "media/381feef28fd8be651d7c/F2.jpg", "media/381feef28fd8be651d7c/F3.jpg", "media/381feef28fd8be651d7c/F4.jpg", "media/381feef28fd8be651d7c/F5.jpg", "media/381feef28fd8be651d7c/F6.jpg", "media/381feef28fd8be651d7c/F7.jpg", "media/381feef28fd8be651d7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (right hand of a person wearing an orange long-sleeved shirt) and object B (left hand of a person wearing an orange long-sleeved shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "596c7beef14673465e09704da82a324d04df0e5f54e28b1b4cab12b0ad18f7de"}
{"qa_id": "VOST__710_cut_paper__camera_translation_direction__00", "video_id": "VOST__710_cut_paper", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "left", "up", "right", "forward"], "frame_paths": ["media/4cf3de05fd1945909c3f/F0.jpg", "media/4cf3de05fd1945909c3f/F1.jpg", "media/4cf3de05fd1945909c3f/F2.jpg", "media/4cf3de05fd1945909c3f/F3.jpg", "media/4cf3de05fd1945909c3f/F4.jpg", "media/4cf3de05fd1945909c3f/F5.jpg", "media/4cf3de05fd1945909c3f/F6.jpg", "media/4cf3de05fd1945909c3f/F7.jpg", "media/4cf3de05fd1945909c3f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. left\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d141d59b594171c031f1d76ad28475b2289629e1c82a05223a636b7e8101c85"}
{"qa_id": "VOST__710_cut_paper__object_pair_distance_change__01", "video_id": "VOST__710_cut_paper", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did the 3D distance between object A (person wearing a black and white checkered shirt, holding scissors) and object B (white paper being cut with scissors) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f8ac1b6eb276b73ba502/F0.jpg", "media/f8ac1b6eb276b73ba502/F1.jpg", "media/f8ac1b6eb276b73ba502/F2.jpg", "media/f8ac1b6eb276b73ba502/F3.jpg", "media/f8ac1b6eb276b73ba502/F4.jpg", "media/f8ac1b6eb276b73ba502/F5.jpg", "media/f8ac1b6eb276b73ba502/F6.jpg", "media/f8ac1b6eb276b73ba502/F7.jpg", "media/f8ac1b6eb276b73ba502/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (person wearing a black and white checkered shirt, holding scissors) and object B (white paper being cut with scissors) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ef8a7e56ca8d4ae0c83484ee2a5c534fe32223cca2e90e0ee420a421db5a9499"}
{"qa_id": "VOST__710_cut_paper__object_travel_distance_comparison__02", "video_id": "VOST__710_cut_paper", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a black and white checkered shirt, holding scissors) or B (white paper being cut with scissors)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f8ac1b6eb276b73ba502/F0.jpg", "media/f8ac1b6eb276b73ba502/F1.jpg", "media/f8ac1b6eb276b73ba502/F2.jpg", "media/f8ac1b6eb276b73ba502/F3.jpg", "media/f8ac1b6eb276b73ba502/F4.jpg", "media/f8ac1b6eb276b73ba502/F5.jpg", "media/f8ac1b6eb276b73ba502/F6.jpg", "media/f8ac1b6eb276b73ba502/F7.jpg", "media/f8ac1b6eb276b73ba502/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a black and white checkered shirt, holding scissors) or B (white paper being cut with scissors)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f36dad46526239c1f6808a739b5d8ff29650bf90c0cfa20276c7e6e14b7a357f"}
{"qa_id": "VOST__718_cut_pepper__object_world_motion_direction__00", "video_id": "VOST__718_cut_pepper", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person wearing a black and white checkered top, cutting a yellow pepper) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "left", "backward", "forward", "right"], "frame_paths": ["media/ed0e4730f8379b5d5525/F0.jpg", "media/ed0e4730f8379b5d5525/F1.jpg", "media/ed0e4730f8379b5d5525/F2.jpg", "media/ed0e4730f8379b5d5525/F3.jpg", "media/ed0e4730f8379b5d5525/F4.jpg", "media/ed0e4730f8379b5d5525/F5.jpg", "media/ed0e4730f8379b5d5525/F6.jpg", "media/ed0e4730f8379b5d5525/F7.jpg", "media/ed0e4730f8379b5d5525/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person wearing a black and white checkered top, cutting a yellow pepper) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "794c04824fd8ed9089234c78ab3fbfdbc38eb85d9f3b2c865f9e756ba1278911"}
{"qa_id": "VOST__7202_clean_car__object_pair_distance_change__00", "video_id": "VOST__7202_clean_car", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did the 3D distance between object A (right arm and hand wearing a black watch, reaching towards a vacuum cleaner wand) and object B (left arm and hand resting on a grey plaid car seat) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/be1f6dac11ee258361f4/F0.jpg", "media/be1f6dac11ee258361f4/F1.jpg", "media/be1f6dac11ee258361f4/F2.jpg", "media/be1f6dac11ee258361f4/F3.jpg", "media/be1f6dac11ee258361f4/F4.jpg", "media/be1f6dac11ee258361f4/F5.jpg", "media/be1f6dac11ee258361f4/F6.jpg", "media/be1f6dac11ee258361f4/F7.jpg", "media/be1f6dac11ee258361f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (right arm and hand wearing a black watch, reaching towards a vacuum cleaner wand) and object B (left arm and hand resting on a grey plaid car seat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d8d152d429bbf61af6bc74166d4b6434403aa27753559c8090c3604520e2949"}
{"qa_id": "VOST__7202_clean_car__object_travel_distance_comparison__01", "video_id": "VOST__7202_clean_car", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right arm and hand wearing a black watch, reaching towards a vacuum cleaner wand) or B (left arm and hand resting on a grey plaid car seat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/be1f6dac11ee258361f4/F0.jpg", "media/be1f6dac11ee258361f4/F1.jpg", "media/be1f6dac11ee258361f4/F2.jpg", "media/be1f6dac11ee258361f4/F3.jpg", "media/be1f6dac11ee258361f4/F4.jpg", "media/be1f6dac11ee258361f4/F5.jpg", "media/be1f6dac11ee258361f4/F6.jpg", "media/be1f6dac11ee258361f4/F7.jpg", "media/be1f6dac11ee258361f4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (right arm and hand wearing a black watch, reaching towards a vacuum cleaner wand) or B (left arm and hand resting on a grey plaid car seat)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "37ebd67c55952e76698572f3bd719378f79a0e2fe7503f6ebaa4b3217df101ff"}
{"qa_id": "VOST__724_cut_cake__camera_translation_direction__00", "video_id": "VOST__724_cut_cake", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "left", "up", "backward"], "frame_paths": ["media/df0b422d218b02559b21/F0.jpg", "media/df0b422d218b02559b21/F1.jpg", "media/df0b422d218b02559b21/F2.jpg", "media/df0b422d218b02559b21/F3.jpg", "media/df0b422d218b02559b21/F4.jpg", "media/df0b422d218b02559b21/F5.jpg", "media/df0b422d218b02559b21/F6.jpg", "media/df0b422d218b02559b21/F7.jpg", "media/df0b422d218b02559b21/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f13a164f986569d0a9f58c16dba80b8c4cb30e023af2168d28ba1d04134a2868"}
{"qa_id": "VOST__7359_fold_tape_measure__camera_translation_direction__00", "video_id": "VOST__7359_fold_tape_measure", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "backward", "forward", "up"], "frame_paths": ["media/7b6a739e81c0913c04c2/F0.jpg", "media/7b6a739e81c0913c04c2/F1.jpg", "media/7b6a739e81c0913c04c2/F2.jpg", "media/7b6a739e81c0913c04c2/F3.jpg", "media/7b6a739e81c0913c04c2/F4.jpg", "media/7b6a739e81c0913c04c2/F5.jpg", "media/7b6a739e81c0913c04c2/F6.jpg", "media/7b6a739e81c0913c04c2/F7.jpg", "media/7b6a739e81c0913c04c2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "21fe0136f2020af3aeff3d9fe20991db484fe48881e729869fc889987268790d"}
{"qa_id": "VOST__739_cut_onion__camera_translation_direction__00", "video_id": "VOST__739_cut_onion", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "forward", "up"], "frame_paths": ["media/38d1b01439b93cc47e6f/F0.jpg", "media/38d1b01439b93cc47e6f/F1.jpg", "media/38d1b01439b93cc47e6f/F2.jpg", "media/38d1b01439b93cc47e6f/F3.jpg", "media/38d1b01439b93cc47e6f/F4.jpg", "media/38d1b01439b93cc47e6f/F5.jpg", "media/38d1b01439b93cc47e6f/F6.jpg", "media/38d1b01439b93cc47e6f/F7.jpg", "media/38d1b01439b93cc47e6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "733998aad2b75384830bebc6020ae006c9c7551a981bd1179da28fa55d08c90c"}
{"qa_id": "VOST__7623_saw_wood__camera_translation_direction__00", "video_id": "VOST__7623_saw_wood", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "backward", "left"], "frame_paths": ["media/e3520661413f38cf27dd/F0.jpg", "media/e3520661413f38cf27dd/F1.jpg", "media/e3520661413f38cf27dd/F2.jpg", "media/e3520661413f38cf27dd/F3.jpg", "media/e3520661413f38cf27dd/F4.jpg", "media/e3520661413f38cf27dd/F5.jpg", "media/e3520661413f38cf27dd/F6.jpg", "media/e3520661413f38cf27dd/F7.jpg", "media/e3520661413f38cf27dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d5fc8ac56aebd0027e79796205b2c7634a0742b480ba2bddf85d194e472eecb"}
{"qa_id": "VOST__7786_break_puzzle__camera_translation_direction__00", "video_id": "VOST__7786_break_puzzle", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "down", "backward", "left"], "frame_paths": ["media/009782d3dbd523995202/F0.jpg", "media/009782d3dbd523995202/F1.jpg", "media/009782d3dbd523995202/F2.jpg", "media/009782d3dbd523995202/F3.jpg", "media/009782d3dbd523995202/F4.jpg", "media/009782d3dbd523995202/F5.jpg", "media/009782d3dbd523995202/F6.jpg", "media/009782d3dbd523995202/F7.jpg", "media/009782d3dbd523995202/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. down\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "00ea6d2133486dc76f0ee24c132cc77ad42ce6578f4fce33931779573bf46b95"}
{"qa_id": "VOST__801_cut_chicken__camera_translation_direction__00", "video_id": "VOST__801_cut_chicken", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "forward", "left", "right", "down"], "frame_paths": ["media/69b3183fc434c794ff6b/F0.jpg", "media/69b3183fc434c794ff6b/F1.jpg", "media/69b3183fc434c794ff6b/F2.jpg", "media/69b3183fc434c794ff6b/F3.jpg", "media/69b3183fc434c794ff6b/F4.jpg", "media/69b3183fc434c794ff6b/F5.jpg", "media/69b3183fc434c794ff6b/F6.jpg", "media/69b3183fc434c794ff6b/F7.jpg", "media/69b3183fc434c794ff6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. forward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "72e3a78ca45cb4fbe6917bb71c771acc9f168ed9ef61f263e98dadb4214658f3"}
{"qa_id": "VOST__801_cut_chicken__camera_view_direction_change__01", "video_id": "VOST__801_cut_chicken", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/69b3183fc434c794ff6b/F0.jpg", "media/69b3183fc434c794ff6b/F1.jpg", "media/69b3183fc434c794ff6b/F2.jpg", "media/69b3183fc434c794ff6b/F3.jpg", "media/69b3183fc434c794ff6b/F4.jpg", "media/69b3183fc434c794ff6b/F5.jpg", "media/69b3183fc434c794ff6b/F6.jpg", "media/69b3183fc434c794ff6b/F7.jpg", "media/69b3183fc434c794ff6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f8424c0b08354e78a0af2158ce3016d19c692a05492f95ce15f4393301f02e3"}
{"qa_id": "VOST__801_cut_chicken__object_world_motion_direction__02", "video_id": "VOST__801_cut_chicken", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (person's left arm) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/ccd6c5f2bc9ea0c721ef/F0.jpg", "media/ccd6c5f2bc9ea0c721ef/F1.jpg", "media/ccd6c5f2bc9ea0c721ef/F2.jpg", "media/ccd6c5f2bc9ea0c721ef/F3.jpg", "media/ccd6c5f2bc9ea0c721ef/F4.jpg", "media/ccd6c5f2bc9ea0c721ef/F5.jpg", "media/ccd6c5f2bc9ea0c721ef/F6.jpg", "media/ccd6c5f2bc9ea0c721ef/F7.jpg", "media/ccd6c5f2bc9ea0c721ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (person's left arm) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7d853184b4adf3da41f7ab43a3ed00318427bef522c7004fb59e8efb57813bef"}
{"qa_id": "VOST__810_cut_dough__camera_translation_direction__00", "video_id": "VOST__810_cut_dough", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "left", "up", "right", "forward"], "frame_paths": ["media/b4efb9cb3d9a7c203859/F0.jpg", "media/b4efb9cb3d9a7c203859/F1.jpg", "media/b4efb9cb3d9a7c203859/F2.jpg", "media/b4efb9cb3d9a7c203859/F3.jpg", "media/b4efb9cb3d9a7c203859/F4.jpg", "media/b4efb9cb3d9a7c203859/F5.jpg", "media/b4efb9cb3d9a7c203859/F6.jpg", "media/b4efb9cb3d9a7c203859/F7.jpg", "media/b4efb9cb3d9a7c203859/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. left\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f68f1a8e0584411789ea59c91f651f3ff3b02083fe1ca91bb2e63631666f235"}
{"qa_id": "VOST__810_cut_dough__camera_view_direction_change__01", "video_id": "VOST__810_cut_dough", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/b4efb9cb3d9a7c203859/F0.jpg", "media/b4efb9cb3d9a7c203859/F1.jpg", "media/b4efb9cb3d9a7c203859/F2.jpg", "media/b4efb9cb3d9a7c203859/F3.jpg", "media/b4efb9cb3d9a7c203859/F4.jpg", "media/b4efb9cb3d9a7c203859/F5.jpg", "media/b4efb9cb3d9a7c203859/F6.jpg", "media/b4efb9cb3d9a7c203859/F7.jpg", "media/b4efb9cb3d9a7c203859/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "VOST__8267_pour_flour__object_pair_distance_change__00", "video_id": "VOST__8267_pour_flour", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, did the 3D distance between object A (white and orange flour bag labeled 'La Spiga') and object B (person's arm wearing a blue glove) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/6463dd12e6e1f4ae8b20/F0.jpg", "media/6463dd12e6e1f4ae8b20/F1.jpg", "media/6463dd12e6e1f4ae8b20/F2.jpg", "media/6463dd12e6e1f4ae8b20/F3.jpg", "media/6463dd12e6e1f4ae8b20/F4.jpg", "media/6463dd12e6e1f4ae8b20/F5.jpg", "media/6463dd12e6e1f4ae8b20/F6.jpg", "media/6463dd12e6e1f4ae8b20/F7.jpg", "media/6463dd12e6e1f4ae8b20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (white and orange flour bag labeled 'La Spiga') and object B (person's arm wearing a blue glove) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3e2defff0136d1693c526307d2f3cb681edae2175ac01bb00f7ba7a076fe59d"}
{"qa_id": "VOST__8560_peel_carrot__camera_view_direction_change__00", "video_id": "VOST__8560_peel_carrot", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/3946ba2faf94f91128f2/F0.jpg", "media/3946ba2faf94f91128f2/F1.jpg", "media/3946ba2faf94f91128f2/F2.jpg", "media/3946ba2faf94f91128f2/F3.jpg", "media/3946ba2faf94f91128f2/F4.jpg", "media/3946ba2faf94f91128f2/F5.jpg", "media/3946ba2faf94f91128f2/F6.jpg", "media/3946ba2faf94f91128f2/F7.jpg", "media/3946ba2faf94f91128f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8b82b5599bf4d761cf41633760b193d10ff47a57ba9610b17d440897b00b84c5"}
{"qa_id": "VOST__8601_peel_ginger__camera_translation_direction__00", "video_id": "VOST__8601_peel_ginger", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "backward", "up", "left"], "frame_paths": ["media/8a4c34fc858831d5880b/F0.jpg", "media/8a4c34fc858831d5880b/F1.jpg", "media/8a4c34fc858831d5880b/F2.jpg", "media/8a4c34fc858831d5880b/F3.jpg", "media/8a4c34fc858831d5880b/F4.jpg", "media/8a4c34fc858831d5880b/F5.jpg", "media/8a4c34fc858831d5880b/F6.jpg", "media/8a4c34fc858831d5880b/F7.jpg", "media/8a4c34fc858831d5880b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5174a6e21aea25b0b313c57db39544926be0a35462b5d565d1117af7aa013f1a"}
{"qa_id": "VOST__8679_roll_dough__camera_translation_direction__00", "video_id": "VOST__8679_roll_dough", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "right", "down", "backward"], "frame_paths": ["media/821c01f93469488cfdce/F0.jpg", "media/821c01f93469488cfdce/F1.jpg", "media/821c01f93469488cfdce/F2.jpg", "media/821c01f93469488cfdce/F3.jpg", "media/821c01f93469488cfdce/F4.jpg", "media/821c01f93469488cfdce/F5.jpg", "media/821c01f93469488cfdce/F6.jpg", "media/821c01f93469488cfdce/F7.jpg", "media/821c01f93469488cfdce/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "37a0e46779d623a1b00f7246a800e2c217d7920bec8da6f8ac5507f3a2746699"}
{"qa_id": "VOST__8918_eat_food__camera_translation_direction__00", "video_id": "VOST__8918_eat_food", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "up", "down", "left"], "frame_paths": ["media/ca2ff1af055717dde144/F0.jpg", "media/ca2ff1af055717dde144/F1.jpg", "media/ca2ff1af055717dde144/F2.jpg", "media/ca2ff1af055717dde144/F3.jpg", "media/ca2ff1af055717dde144/F4.jpg", "media/ca2ff1af055717dde144/F5.jpg", "media/ca2ff1af055717dde144/F6.jpg", "media/ca2ff1af055717dde144/F7.jpg", "media/ca2ff1af055717dde144/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc3e54386240133079e541c474a24fa56c9fb68572661ddc3f4ba8ff7ba8eee7"}
{"qa_id": "VOST__8918_eat_food__camera_object_distance_change__01", "video_id": "VOST__8918_eat_food", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did object A (small black rectangular device with a screen displaying a purple light) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a0723e7c67e6276e6eb5/F0.jpg", "media/a0723e7c67e6276e6eb5/F1.jpg", "media/a0723e7c67e6276e6eb5/F2.jpg", "media/a0723e7c67e6276e6eb5/F3.jpg", "media/a0723e7c67e6276e6eb5/F4.jpg", "media/a0723e7c67e6276e6eb5/F5.jpg", "media/a0723e7c67e6276e6eb5/F6.jpg", "media/a0723e7c67e6276e6eb5/F7.jpg", "media/a0723e7c67e6276e6eb5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (small black rectangular device with a screen displaying a purple light) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0fac00b371e28c1d50c7454fe85417821bea5aba44b66afb4e11f3060ac8d38a"}
{"qa_id": "VOST__8921_eat_food__camera_translation_direction__00", "video_id": "VOST__8921_eat_food", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "down", "up", "left", "forward", "backward"], "frame_paths": ["media/47f6f9669142c50323ba/F0.jpg", "media/47f6f9669142c50323ba/F1.jpg", "media/47f6f9669142c50323ba/F2.jpg", "media/47f6f9669142c50323ba/F3.jpg", "media/47f6f9669142c50323ba/F4.jpg", "media/47f6f9669142c50323ba/F5.jpg", "media/47f6f9669142c50323ba/F6.jpg", "media/47f6f9669142c50323ba/F7.jpg", "media/47f6f9669142c50323ba/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b0406e70264604504ac53ae3c8c83a46160a23c90645be251d22bcbd642a1f51"}
{"qa_id": "VOST__9119_break_nut__camera_translation_direction__00", "video_id": "VOST__9119_break_nut", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "forward", "left", "right"], "frame_paths": ["media/619468e86ad77d6a9207/F0.jpg", "media/619468e86ad77d6a9207/F1.jpg", "media/619468e86ad77d6a9207/F2.jpg", "media/619468e86ad77d6a9207/F3.jpg", "media/619468e86ad77d6a9207/F4.jpg", "media/619468e86ad77d6a9207/F5.jpg", "media/619468e86ad77d6a9207/F6.jpg", "media/619468e86ad77d6a9207/F7.jpg", "media/619468e86ad77d6a9207/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44d0e205ae2c57f23abd627b516d93539dc3721e4ce25b84e6ad3b0fe9f70d20"}
{"qa_id": "VOST__9258_split_cards__camera_translation_direction__00", "video_id": "VOST__9258_split_cards", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "down", "left", "up"], "frame_paths": ["media/00477664ac39b28308a0/F0.jpg", "media/00477664ac39b28308a0/F1.jpg", "media/00477664ac39b28308a0/F2.jpg", "media/00477664ac39b28308a0/F3.jpg", "media/00477664ac39b28308a0/F4.jpg", "media/00477664ac39b28308a0/F5.jpg", "media/00477664ac39b28308a0/F6.jpg", "media/00477664ac39b28308a0/F7.jpg", "media/00477664ac39b28308a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f30a32d875c91ed87d11e38d6c6eb0a220b7361fd59107cda132f4f1f7a7b700"}
{"qa_id": "VOST__9285_apply_paint__object_world_motion_direction__00", "video_id": "VOST__9285_apply_paint", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (person's right arm and hand holding a paintbrush) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "left", "up", "down", "forward"], "frame_paths": ["media/9018cb2a76030acee4e5/F0.jpg", "media/9018cb2a76030acee4e5/F1.jpg", "media/9018cb2a76030acee4e5/F2.jpg", "media/9018cb2a76030acee4e5/F3.jpg", "media/9018cb2a76030acee4e5/F4.jpg", "media/9018cb2a76030acee4e5/F5.jpg", "media/9018cb2a76030acee4e5/F6.jpg", "media/9018cb2a76030acee4e5/F7.jpg", "media/9018cb2a76030acee4e5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (person's right arm and hand holding a paintbrush) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. left\nD. up\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbf0a63019cf3ff5a00ce34740ac35cb68acc22c77d9e10216c7df5745eabf83"}
{"qa_id": "VOST__9285_apply_paint__object_pair_distance_change__01", "video_id": "VOST__9285_apply_paint", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did the 3D distance between object A (person's right arm and hand holding a paintbrush) and object B (person's left arm wearing a grey patterned sleeve) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/ef68965159d71e1179e7/F0.jpg", "media/ef68965159d71e1179e7/F1.jpg", "media/ef68965159d71e1179e7/F2.jpg", "media/ef68965159d71e1179e7/F3.jpg", "media/ef68965159d71e1179e7/F4.jpg", "media/ef68965159d71e1179e7/F5.jpg", "media/ef68965159d71e1179e7/F6.jpg", "media/ef68965159d71e1179e7/F7.jpg", "media/ef68965159d71e1179e7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (person's right arm and hand holding a paintbrush) and object B (person's left arm wearing a grey patterned sleeve) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38a4be5ebea194694f0c977f523525f0321b1070ff50f1aec2b9c0280e895abf"}
{"qa_id": "VOST__9285_apply_paint__object_travel_distance_comparison__02", "video_id": "VOST__9285_apply_paint", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a paintbrush) or B (person's left arm wearing a grey patterned sleeve)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/ef68965159d71e1179e7/F0.jpg", "media/ef68965159d71e1179e7/F1.jpg", "media/ef68965159d71e1179e7/F2.jpg", "media/ef68965159d71e1179e7/F3.jpg", "media/ef68965159d71e1179e7/F4.jpg", "media/ef68965159d71e1179e7/F5.jpg", "media/ef68965159d71e1179e7/F6.jpg", "media/ef68965159d71e1179e7/F7.jpg", "media/ef68965159d71e1179e7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's right arm and hand holding a paintbrush) or B (person's left arm wearing a grey patterned sleeve)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b7b982056404c113a0380996da845107f51c40cfe7b096e9646b10384a383df"}
{"qa_id": "VOST__928_cut_chicken__camera_translation_direction__00", "video_id": "VOST__928_cut_chicken", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "right", "up", "down", "left"], "frame_paths": ["media/85144ea3a197f9fbffe7/F0.jpg", "media/85144ea3a197f9fbffe7/F1.jpg", "media/85144ea3a197f9fbffe7/F2.jpg", "media/85144ea3a197f9fbffe7/F3.jpg", "media/85144ea3a197f9fbffe7/F4.jpg", "media/85144ea3a197f9fbffe7/F5.jpg", "media/85144ea3a197f9fbffe7/F6.jpg", "media/85144ea3a197f9fbffe7/F7.jpg", "media/85144ea3a197f9fbffe7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc7a5b7fcc3b0eeddbb57d478595563f0e312eb9e8e872341e8c2db6f4cb1b29"}
{"qa_id": "VOST__928_cut_chicken__object_travel_distance_comparison__01", "video_id": "VOST__928_cut_chicken", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm in a blue sleeve holding food on a cutting board) or B (person's right arm in a blue sleeve holding a knife)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/10e18f42dad133dfc8a0/F0.jpg", "media/10e18f42dad133dfc8a0/F1.jpg", "media/10e18f42dad133dfc8a0/F2.jpg", "media/10e18f42dad133dfc8a0/F3.jpg", "media/10e18f42dad133dfc8a0/F4.jpg", "media/10e18f42dad133dfc8a0/F5.jpg", "media/10e18f42dad133dfc8a0/F6.jpg", "media/10e18f42dad133dfc8a0/F7.jpg", "media/10e18f42dad133dfc8a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm in a blue sleeve holding food on a cutting board) or B (person's right arm in a blue sleeve holding a knife)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "532bce2aa079d1ad586e595fe27f04d6b20a2c0b43934517ff1206146b1f8c9f"}
{"qa_id": "VOST__9291_apply_paint__camera_translation_direction__00", "video_id": "VOST__9291_apply_paint", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "forward", "right", "down"], "frame_paths": ["media/93fac26bf16f783cf68f/F0.jpg", "media/93fac26bf16f783cf68f/F1.jpg", "media/93fac26bf16f783cf68f/F2.jpg", "media/93fac26bf16f783cf68f/F3.jpg", "media/93fac26bf16f783cf68f/F4.jpg", "media/93fac26bf16f783cf68f/F5.jpg", "media/93fac26bf16f783cf68f/F6.jpg", "media/93fac26bf16f783cf68f/F7.jpg", "media/93fac26bf16f783cf68f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8fc70d539c12ad38cd44534ba5c81962bb5268a2cc6921a87cc24f857112613e"}
{"qa_id": "VOST__9366_pour_batter__camera_translation_direction__00", "video_id": "VOST__9366_pour_batter", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "up", "backward", "right"], "frame_paths": ["media/6a1935dd35d3906540bd/F0.jpg", "media/6a1935dd35d3906540bd/F1.jpg", "media/6a1935dd35d3906540bd/F2.jpg", "media/6a1935dd35d3906540bd/F3.jpg", "media/6a1935dd35d3906540bd/F4.jpg", "media/6a1935dd35d3906540bd/F5.jpg", "media/6a1935dd35d3906540bd/F6.jpg", "media/6a1935dd35d3906540bd/F7.jpg", "media/6a1935dd35d3906540bd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dd86ae169feedb58ce1a13f273a60ed82b281f0ccff8cedaf3a14104e6073ca8"}
{"qa_id": "VOST__9368_pour_batter__object_world_motion_direction__00", "video_id": "VOST__9368_pour_batter", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object A (bare arm reaching towards a stove) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "right", "forward", "left"], "frame_paths": ["media/b3681f87ab432e682820/F0.jpg", "media/b3681f87ab432e682820/F1.jpg", "media/b3681f87ab432e682820/F2.jpg", "media/b3681f87ab432e682820/F3.jpg", "media/b3681f87ab432e682820/F4.jpg", "media/b3681f87ab432e682820/F5.jpg", "media/b3681f87ab432e682820/F6.jpg", "media/b3681f87ab432e682820/F7.jpg", "media/b3681f87ab432e682820/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object A (bare arm reaching towards a stove) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe419f2accd2e1f05bc4d74071e5f69bd4a717d2a451098ad9ecf1c94853e659"}
{"qa_id": "VOST__9368_pour_batter__object_pair_distance_change__01", "video_id": "VOST__9368_pour_batter", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did the 3D distance between object A (bare arm reaching towards a stove) and object B (metal spoon with a light-colored handle) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/6cffd5ab7547c8fb844b/F0.jpg", "media/6cffd5ab7547c8fb844b/F1.jpg", "media/6cffd5ab7547c8fb844b/F2.jpg", "media/6cffd5ab7547c8fb844b/F3.jpg", "media/6cffd5ab7547c8fb844b/F4.jpg", "media/6cffd5ab7547c8fb844b/F5.jpg", "media/6cffd5ab7547c8fb844b/F6.jpg", "media/6cffd5ab7547c8fb844b/F7.jpg", "media/6cffd5ab7547c8fb844b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (bare arm reaching towards a stove) and object B (metal spoon with a light-colored handle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a6a7d430dce3838477684444dc7c2f99e46fdeb642d6c9affb992b176fbd2db"}
{"qa_id": "VOST__9369_scoop_batter__object_world_motion_direction__00", "video_id": "VOST__9369_scoop_batter", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (bare arm reaching towards a pot on the stove) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "down", "right", "left"], "frame_paths": ["media/6bbb7e05b337d8f927a2/F0.jpg", "media/6bbb7e05b337d8f927a2/F1.jpg", "media/6bbb7e05b337d8f927a2/F2.jpg", "media/6bbb7e05b337d8f927a2/F3.jpg", "media/6bbb7e05b337d8f927a2/F4.jpg", "media/6bbb7e05b337d8f927a2/F5.jpg", "media/6bbb7e05b337d8f927a2/F6.jpg", "media/6bbb7e05b337d8f927a2/F7.jpg", "media/6bbb7e05b337d8f927a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (bare arm reaching towards a pot on the stove) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "175411db9808ef918a477e1ebfa9fd24070e57785e79464a934aec104d16b392"}
{"qa_id": "VOST__9369_scoop_batter__object_travel_distance_comparison__01", "video_id": "VOST__9369_scoop_batter", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (bare arm reaching towards a pot on the stove) or B (upper torso and shoulder of a person)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/d4382c9f6d55056e26f5/F0.jpg", "media/d4382c9f6d55056e26f5/F1.jpg", "media/d4382c9f6d55056e26f5/F2.jpg", "media/d4382c9f6d55056e26f5/F3.jpg", "media/d4382c9f6d55056e26f5/F4.jpg", "media/d4382c9f6d55056e26f5/F5.jpg", "media/d4382c9f6d55056e26f5/F6.jpg", "media/d4382c9f6d55056e26f5/F7.jpg", "media/d4382c9f6d55056e26f5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (bare arm reaching towards a pot on the stove) or B (upper torso and shoulder of a person)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a0ecb7db46fdd2bdb6fe0fad1d8e4bdd2fe92909e482c76425d94b40fe997ba8"}
{"qa_id": "VOST__9370_scoop_batter__camera_translation_direction__00", "video_id": "VOST__9370_scoop_batter", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/753e499fd55e66ac8cd7/F0.jpg", "media/753e499fd55e66ac8cd7/F1.jpg", "media/753e499fd55e66ac8cd7/F2.jpg", "media/753e499fd55e66ac8cd7/F3.jpg", "media/753e499fd55e66ac8cd7/F4.jpg", "media/753e499fd55e66ac8cd7/F5.jpg", "media/753e499fd55e66ac8cd7/F6.jpg", "media/753e499fd55e66ac8cd7/F7.jpg", "media/753e499fd55e66ac8cd7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5a7600045a78368e27c66685b2f35d6e0a8a381470bbe1063cbdb206f47c831"}
{"qa_id": "VOST__9370_scoop_batter__object_world_motion_direction__01", "video_id": "VOST__9370_scoop_batter", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (arm reaching towards a pan on the stove) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "right", "up", "down", "backward"], "frame_paths": ["media/5035c3e761abd4bae05e/F0.jpg", "media/5035c3e761abd4bae05e/F1.jpg", "media/5035c3e761abd4bae05e/F2.jpg", "media/5035c3e761abd4bae05e/F3.jpg", "media/5035c3e761abd4bae05e/F4.jpg", "media/5035c3e761abd4bae05e/F5.jpg", "media/5035c3e761abd4bae05e/F6.jpg", "media/5035c3e761abd4bae05e/F7.jpg", "media/5035c3e761abd4bae05e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (arm reaching towards a pan on the stove) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. right\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a48aedc54394e09d10585ae35f7c67481f63aba16d7294cb3341f1cf27941223"}
{"qa_id": "VOST__9370_scoop_batter__object_travel_distance_comparison__02", "video_id": "VOST__9370_scoop_batter", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (arm reaching towards a pan on the stove) or B (torso of a person wearing a dark shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/ee64186ae1b8af149d22/F0.jpg", "media/ee64186ae1b8af149d22/F1.jpg", "media/ee64186ae1b8af149d22/F2.jpg", "media/ee64186ae1b8af149d22/F3.jpg", "media/ee64186ae1b8af149d22/F4.jpg", "media/ee64186ae1b8af149d22/F5.jpg", "media/ee64186ae1b8af149d22/F6.jpg", "media/ee64186ae1b8af149d22/F7.jpg", "media/ee64186ae1b8af149d22/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (arm reaching towards a pan on the stove) or B (torso of a person wearing a dark shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34d6e5cfc0b277cd10e25fa5f27e79fd7339616b334fb089d4a8aba1a8b69e43"}
{"qa_id": "VOST__9398_scoop_cement__camera_translation_direction__00", "video_id": "VOST__9398_scoop_cement", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/95786b34d0195cd9c716/F0.jpg", "media/95786b34d0195cd9c716/F1.jpg", "media/95786b34d0195cd9c716/F2.jpg", "media/95786b34d0195cd9c716/F3.jpg", "media/95786b34d0195cd9c716/F4.jpg", "media/95786b34d0195cd9c716/F5.jpg", "media/95786b34d0195cd9c716/F6.jpg", "media/95786b34d0195cd9c716/F7.jpg", "media/95786b34d0195cd9c716/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b55319fb2710244a8c4c61c69f41eb900276ecb8a9f3c9e14f515b79ec035b37"}
{"qa_id": "VOST__9398_scoop_cement__camera_view_direction_change__01", "video_id": "VOST__9398_scoop_cement", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/95786b34d0195cd9c716/F0.jpg", "media/95786b34d0195cd9c716/F1.jpg", "media/95786b34d0195cd9c716/F2.jpg", "media/95786b34d0195cd9c716/F3.jpg", "media/95786b34d0195cd9c716/F4.jpg", "media/95786b34d0195cd9c716/F5.jpg", "media/95786b34d0195cd9c716/F6.jpg", "media/95786b34d0195cd9c716/F7.jpg", "media/95786b34d0195cd9c716/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ac570cf6b336203bd258e90fd80f7b38dabeda9aba416b5a17cca5c167a9fed2"}
{"qa_id": "VOST__9688_divide_dough__camera_translation_direction__00", "video_id": "VOST__9688_divide_dough", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "forward", "down", "right"], "frame_paths": ["media/6565c48437bfd0986e6f/F0.jpg", "media/6565c48437bfd0986e6f/F1.jpg", "media/6565c48437bfd0986e6f/F2.jpg", "media/6565c48437bfd0986e6f/F3.jpg", "media/6565c48437bfd0986e6f/F4.jpg", "media/6565c48437bfd0986e6f/F5.jpg", "media/6565c48437bfd0986e6f/F6.jpg", "media/6565c48437bfd0986e6f/F7.jpg", "media/6565c48437bfd0986e6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. forward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0fc0a2e7aa5eb487a3bdf9e2f5960b18bab92ced6904bad37a30960f838444f0"}
{"qa_id": "VOST__971_cut_onion__camera_translation_direction__00", "video_id": "VOST__971_cut_onion", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "backward", "down", "up"], "frame_paths": ["media/c3d50c0266c510c9a341/F0.jpg", "media/c3d50c0266c510c9a341/F1.jpg", "media/c3d50c0266c510c9a341/F2.jpg", "media/c3d50c0266c510c9a341/F3.jpg", "media/c3d50c0266c510c9a341/F4.jpg", "media/c3d50c0266c510c9a341/F5.jpg", "media/c3d50c0266c510c9a341/F6.jpg", "media/c3d50c0266c510c9a341/F7.jpg", "media/c3d50c0266c510c9a341/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e33e0e89f869e89bfc97167844d0fcb52b908b6d5b97d120f0abd51e382266b6"}
{"qa_id": "VOST__971_cut_onion__object_world_motion_direction__01", "video_id": "VOST__971_cut_onion", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person's left arm and hand holding a knife, wearing a watch) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "right", "down", "forward"], "frame_paths": ["media/b400a2ca80a2fc05525f/F0.jpg", "media/b400a2ca80a2fc05525f/F1.jpg", "media/b400a2ca80a2fc05525f/F2.jpg", "media/b400a2ca80a2fc05525f/F3.jpg", "media/b400a2ca80a2fc05525f/F4.jpg", "media/b400a2ca80a2fc05525f/F5.jpg", "media/b400a2ca80a2fc05525f/F6.jpg", "media/b400a2ca80a2fc05525f/F7.jpg", "media/b400a2ca80a2fc05525f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person's left arm and hand holding a knife, wearing a watch) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c247410b1b9b2607f725e16656fba266ab8d142b46d56c8aa83b9efe676ffe9c"}
{"qa_id": "VOST__971_cut_onion__object_travel_distance_comparison__02", "video_id": "VOST__971_cut_onion", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand holding a knife, wearing a watch) or B (person's right arm and hand holding food, wearing a grey patterned shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/747e443bc0e8a37a07b6/F0.jpg", "media/747e443bc0e8a37a07b6/F1.jpg", "media/747e443bc0e8a37a07b6/F2.jpg", "media/747e443bc0e8a37a07b6/F3.jpg", "media/747e443bc0e8a37a07b6/F4.jpg", "media/747e443bc0e8a37a07b6/F5.jpg", "media/747e443bc0e8a37a07b6/F6.jpg", "media/747e443bc0e8a37a07b6/F7.jpg", "media/747e443bc0e8a37a07b6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand holding a knife, wearing a watch) or B (person's right arm and hand holding food, wearing a grey patterned shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13267a97db5c7da70914c3ae205b8afe7ab39d2a3917037ee9b91cfc2fbc233a"}
{"qa_id": "VOST__9857_sew_cloth__camera_translation_direction__00", "video_id": "VOST__9857_sew_cloth", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "down", "up", "forward"], "frame_paths": ["media/68c8093d6e4c79cd5bf3/F0.jpg", "media/68c8093d6e4c79cd5bf3/F1.jpg", "media/68c8093d6e4c79cd5bf3/F2.jpg", "media/68c8093d6e4c79cd5bf3/F3.jpg", "media/68c8093d6e4c79cd5bf3/F4.jpg", "media/68c8093d6e4c79cd5bf3/F5.jpg", "media/68c8093d6e4c79cd5bf3/F6.jpg", "media/68c8093d6e4c79cd5bf3/F7.jpg", "media/68c8093d6e4c79cd5bf3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38f3c2ff65364ef0b148331c8b8ee1530846d76889029f51682afde67934b8fb"}
{"qa_id": "VOST__9857_sew_cloth__camera_view_direction_change__01", "video_id": "VOST__9857_sew_cloth", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/68c8093d6e4c79cd5bf3/F0.jpg", "media/68c8093d6e4c79cd5bf3/F1.jpg", "media/68c8093d6e4c79cd5bf3/F2.jpg", "media/68c8093d6e4c79cd5bf3/F3.jpg", "media/68c8093d6e4c79cd5bf3/F4.jpg", "media/68c8093d6e4c79cd5bf3/F5.jpg", "media/68c8093d6e4c79cd5bf3/F6.jpg", "media/68c8093d6e4c79cd5bf3/F7.jpg", "media/68c8093d6e4c79cd5bf3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "daea01c2afd950f5019daa56e0ce06fa0f286d03575f5bdae183e746898299dc"}
{"qa_id": "VOST__994_cut_onion__camera_translation_direction__00", "video_id": "VOST__994_cut_onion", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "right", "backward", "up"], "frame_paths": ["media/8b93274cd6c1d186d184/F0.jpg", "media/8b93274cd6c1d186d184/F1.jpg", "media/8b93274cd6c1d186d184/F2.jpg", "media/8b93274cd6c1d186d184/F3.jpg", "media/8b93274cd6c1d186d184/F4.jpg", "media/8b93274cd6c1d186d184/F5.jpg", "media/8b93274cd6c1d186d184/F6.jpg", "media/8b93274cd6c1d186d184/F7.jpg", "media/8b93274cd6c1d186d184/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2ed54727c21c4c3443de0a0d92f464dc5fc5b753ae3aabc97f4d90d778e76179"}
{"qa_id": "VOST__994_cut_onion__object_world_motion_direction__01", "video_id": "VOST__994_cut_onion", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person wearing a black long-sleeved shirt with white text, reaching towards a plate) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/e7ec1845d5c0bca7d62c/F0.jpg", "media/e7ec1845d5c0bca7d62c/F1.jpg", "media/e7ec1845d5c0bca7d62c/F2.jpg", "media/e7ec1845d5c0bca7d62c/F3.jpg", "media/e7ec1845d5c0bca7d62c/F4.jpg", "media/e7ec1845d5c0bca7d62c/F5.jpg", "media/e7ec1845d5c0bca7d62c/F6.jpg", "media/e7ec1845d5c0bca7d62c/F7.jpg", "media/e7ec1845d5c0bca7d62c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person wearing a black long-sleeved shirt with white text, reaching towards a plate) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e4e9cf86aa5a452f0beefd1f53457cb6bf732a8d4d74494751e18b9468353ce"}
{"qa_id": "VOST__996_cut_bacon__camera_translation_direction__00", "video_id": "VOST__996_cut_bacon", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "up", "down", "forward"], "frame_paths": ["media/3e0492e4d4e840d6aab8/F0.jpg", "media/3e0492e4d4e840d6aab8/F1.jpg", "media/3e0492e4d4e840d6aab8/F2.jpg", "media/3e0492e4d4e840d6aab8/F3.jpg", "media/3e0492e4d4e840d6aab8/F4.jpg", "media/3e0492e4d4e840d6aab8/F5.jpg", "media/3e0492e4d4e840d6aab8/F6.jpg", "media/3e0492e4d4e840d6aab8/F7.jpg", "media/3e0492e4d4e840d6aab8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. up\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b0caf11ffaee06f6af2c8314ddcf8aa7bfa03256fb89bf6f1ccd41f0c76cc988"}
{"qa_id": "VOST__996_cut_bacon__object_travel_distance_comparison__01", "video_id": "VOST__996_cut_bacon", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand wearing a black long-sleeved shirt, reaching towards a plate) or B (person's torso and right hand wearing a black t-shirt with white graphic text, holding a knife)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/8d4a5a49b2f916eb1a7b/F0.jpg", "media/8d4a5a49b2f916eb1a7b/F1.jpg", "media/8d4a5a49b2f916eb1a7b/F2.jpg", "media/8d4a5a49b2f916eb1a7b/F3.jpg", "media/8d4a5a49b2f916eb1a7b/F4.jpg", "media/8d4a5a49b2f916eb1a7b/F5.jpg", "media/8d4a5a49b2f916eb1a7b/F6.jpg", "media/8d4a5a49b2f916eb1a7b/F7.jpg", "media/8d4a5a49b2f916eb1a7b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person's left arm and hand wearing a black long-sleeved shirt, reaching towards a plate) or B (person's torso and right hand wearing a black t-shirt with white graphic text, holding a knife)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "57fb06a87d4a74d7ee25fda0ac5c6d1617994879aa43290fc945c8d356e6d947"}
{"qa_id": "uvo__-2Km-NboGO4__camera_view_direction_change__00", "video_id": "uvo__-2Km-NboGO4", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/a2a03f2d4b3016e926a6/F0.jpg", "media/a2a03f2d4b3016e926a6/F1.jpg", "media/a2a03f2d4b3016e926a6/F2.jpg", "media/a2a03f2d4b3016e926a6/F3.jpg", "media/a2a03f2d4b3016e926a6/F4.jpg", "media/a2a03f2d4b3016e926a6/F5.jpg", "media/a2a03f2d4b3016e926a6/F6.jpg", "media/a2a03f2d4b3016e926a6/F7.jpg", "media/a2a03f2d4b3016e926a6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "628412a7a6a40e3160b77784c8f4351610f72858cb6929cf596ee4afe5c2ce91"}
{"qa_id": "uvo__-327uLLh08w__camera_translation_direction__00", "video_id": "uvo__-327uLLh08w", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "up", "right"], "frame_paths": ["media/3ccf7860a32a42b23d35/F0.jpg", "media/3ccf7860a32a42b23d35/F1.jpg", "media/3ccf7860a32a42b23d35/F2.jpg", "media/3ccf7860a32a42b23d35/F3.jpg", "media/3ccf7860a32a42b23d35/F4.jpg", "media/3ccf7860a32a42b23d35/F5.jpg", "media/3ccf7860a32a42b23d35/F6.jpg", "media/3ccf7860a32a42b23d35/F7.jpg", "media/3ccf7860a32a42b23d35/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b0a8e1d02d3f9570fc6c2b135f66b11474562c9ce20be996b67b6c029766a563"}
{"qa_id": "uvo__-3OVmEFi_X4__object_world_motion_direction__00", "video_id": "uvo__-3OVmEFi_X4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (person standing on the pool deck) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "down", "left", "up"], "frame_paths": ["media/440235b6663edc8a4ddb/F0.jpg", "media/440235b6663edc8a4ddb/F1.jpg", "media/440235b6663edc8a4ddb/F2.jpg", "media/440235b6663edc8a4ddb/F3.jpg", "media/440235b6663edc8a4ddb/F4.jpg", "media/440235b6663edc8a4ddb/F5.jpg", "media/440235b6663edc8a4ddb/F6.jpg", "media/440235b6663edc8a4ddb/F7.jpg", "media/440235b6663edc8a4ddb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (person standing on the pool deck) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e0985c1e3f67ad200f016a9d3c1ae0f24411d66a9cc374dcdf7254616108f88"}
{"qa_id": "uvo__-3OVmEFi_X4__camera_object_distance_change__01", "video_id": "uvo__-3OVmEFi_X4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (person standing on the pool deck) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/440235b6663edc8a4ddb/F0.jpg", "media/440235b6663edc8a4ddb/F1.jpg", "media/440235b6663edc8a4ddb/F2.jpg", "media/440235b6663edc8a4ddb/F3.jpg", "media/440235b6663edc8a4ddb/F4.jpg", "media/440235b6663edc8a4ddb/F5.jpg", "media/440235b6663edc8a4ddb/F6.jpg", "media/440235b6663edc8a4ddb/F7.jpg", "media/440235b6663edc8a4ddb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (person standing on the pool deck) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7798c876e83522eaf4fcdd46b8b4dae3ce78ab57df30237cce0b434e0c2a0979"}
{"qa_id": "uvo__-3OVmEFi_X4__object_pair_distance_change__02", "video_id": "uvo__-3OVmEFi_X4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (person standing on the pool deck) and object B (person standing on the pool deck to the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/38ffa55845c16c2aeae2/F0.jpg", "media/38ffa55845c16c2aeae2/F1.jpg", "media/38ffa55845c16c2aeae2/F2.jpg", "media/38ffa55845c16c2aeae2/F3.jpg", "media/38ffa55845c16c2aeae2/F4.jpg", "media/38ffa55845c16c2aeae2/F5.jpg", "media/38ffa55845c16c2aeae2/F6.jpg", "media/38ffa55845c16c2aeae2/F7.jpg", "media/38ffa55845c16c2aeae2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person standing on the pool deck) and object B (person standing on the pool deck to the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "696d9e9e0f85aae9b834218d53786f5ac20f6a7572c5f0a6ff9e05ae476409e8"}
{"qa_id": "uvo__-3W9JmAcHxI__camera_translation_direction__00", "video_id": "uvo__-3W9JmAcHxI", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "forward", "backward", "left", "down"], "frame_paths": ["media/638de00a6caf12f69c20/F0.jpg", "media/638de00a6caf12f69c20/F1.jpg", "media/638de00a6caf12f69c20/F2.jpg", "media/638de00a6caf12f69c20/F3.jpg", "media/638de00a6caf12f69c20/F4.jpg", "media/638de00a6caf12f69c20/F5.jpg", "media/638de00a6caf12f69c20/F6.jpg", "media/638de00a6caf12f69c20/F7.jpg", "media/638de00a6caf12f69c20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. forward\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c1a2a5273b3f6f1002cb1982edcd91674d4f21e1b9da76d1ed13b2e2599591e5"}
{"qa_id": "uvo__-3W9JmAcHxI__camera_object_distance_change__01", "video_id": "uvo__-3W9JmAcHxI", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, did object A (person in orange jacket and black pants climbing a snowy slope) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f3abbdfcad6505611666/F0.jpg", "media/f3abbdfcad6505611666/F1.jpg", "media/f3abbdfcad6505611666/F2.jpg", "media/f3abbdfcad6505611666/F3.jpg", "media/f3abbdfcad6505611666/F4.jpg", "media/f3abbdfcad6505611666/F5.jpg", "media/f3abbdfcad6505611666/F6.jpg", "media/f3abbdfcad6505611666/F7.jpg", "media/f3abbdfcad6505611666/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (person in orange jacket and black pants climbing a snowy slope) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7d3461e6aef0a9a6d145b01e8d97a00a3bcefe0ae57bd99a46af30bd66dcf11b"}
{"qa_id": "uvo__-3pWBL1TJZg__camera_translation_direction__00", "video_id": "uvo__-3pWBL1TJZg", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "right", "left"], "frame_paths": ["media/09f97bef84fed46f772c/F0.jpg", "media/09f97bef84fed46f772c/F1.jpg", "media/09f97bef84fed46f772c/F2.jpg", "media/09f97bef84fed46f772c/F3.jpg", "media/09f97bef84fed46f772c/F4.jpg", "media/09f97bef84fed46f772c/F5.jpg", "media/09f97bef84fed46f772c/F6.jpg", "media/09f97bef84fed46f772c/F7.jpg", "media/09f97bef84fed46f772c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7fab3fdf84bea373a5597fa71884a7f4887d3104ee5b1a2b306eb6923a623d72"}
{"qa_id": "uvo__-3pWBL1TJZg__camera_object_distance_change__01", "video_id": "uvo__-3pWBL1TJZg", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, did object A (ice climber in red jacket and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d31032c57c33fd211b72/F0.jpg", "media/d31032c57c33fd211b72/F1.jpg", "media/d31032c57c33fd211b72/F2.jpg", "media/d31032c57c33fd211b72/F3.jpg", "media/d31032c57c33fd211b72/F4.jpg", "media/d31032c57c33fd211b72/F5.jpg", "media/d31032c57c33fd211b72/F6.jpg", "media/d31032c57c33fd211b72/F7.jpg", "media/d31032c57c33fd211b72/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (ice climber in red jacket and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "368ec1310d53eabb9ac8281e8aea12df3f5e55c850491662b657a4704ecac292"}
{"qa_id": "uvo__-3wWE2UjjWM__camera_view_direction_change__00", "video_id": "uvo__-3wWE2UjjWM", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/fc7001fdc4d8d5799e7d/F0.jpg", "media/fc7001fdc4d8d5799e7d/F1.jpg", "media/fc7001fdc4d8d5799e7d/F2.jpg", "media/fc7001fdc4d8d5799e7d/F3.jpg", "media/fc7001fdc4d8d5799e7d/F4.jpg", "media/fc7001fdc4d8d5799e7d/F5.jpg", "media/fc7001fdc4d8d5799e7d/F6.jpg", "media/fc7001fdc4d8d5799e7d/F7.jpg", "media/fc7001fdc4d8d5799e7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "uvo__-4_M902vHy8__camera_view_direction_change__00", "video_id": "uvo__-4_M902vHy8", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/6febff50cd6aeb3ee6f8/F0.jpg", "media/6febff50cd6aeb3ee6f8/F1.jpg", "media/6febff50cd6aeb3ee6f8/F2.jpg", "media/6febff50cd6aeb3ee6f8/F3.jpg", "media/6febff50cd6aeb3ee6f8/F4.jpg", "media/6febff50cd6aeb3ee6f8/F5.jpg", "media/6febff50cd6aeb3ee6f8/F6.jpg", "media/6febff50cd6aeb3ee6f8/F7.jpg", "media/6febff50cd6aeb3ee6f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0bceae88ffaa97dfd8969e8ac1654b8a978f5efa8e54671a7323092394a0ae03"}
{"qa_id": "uvo__-4dHxTY4iwo__camera_translation_direction__00", "video_id": "uvo__-4dHxTY4iwo", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "left", "forward", "right"], "frame_paths": ["media/5a9eb3f993cedb0a6c83/F0.jpg", "media/5a9eb3f993cedb0a6c83/F1.jpg", "media/5a9eb3f993cedb0a6c83/F2.jpg", "media/5a9eb3f993cedb0a6c83/F3.jpg", "media/5a9eb3f993cedb0a6c83/F4.jpg", "media/5a9eb3f993cedb0a6c83/F5.jpg", "media/5a9eb3f993cedb0a6c83/F6.jpg", "media/5a9eb3f993cedb0a6c83/F7.jpg", "media/5a9eb3f993cedb0a6c83/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. left\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6788f19e4a23a6896768ff3e1f70d200709598f851f2b6e2dc4daa1bbddee5ef"}
{"qa_id": "uvo__-51p8XCHUvU__camera_view_direction_change__00", "video_id": "uvo__-51p8XCHUvU", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/8ccfc7218a33c3c4194d/F0.jpg", "media/8ccfc7218a33c3c4194d/F1.jpg", "media/8ccfc7218a33c3c4194d/F2.jpg", "media/8ccfc7218a33c3c4194d/F3.jpg", "media/8ccfc7218a33c3c4194d/F4.jpg", "media/8ccfc7218a33c3c4194d/F5.jpg", "media/8ccfc7218a33c3c4194d/F6.jpg", "media/8ccfc7218a33c3c4194d/F7.jpg", "media/8ccfc7218a33c3c4194d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9ea7733431d7e736258a8696695bb4cec708231e380c3fbc6fbf5ac119c2971b"}
{"qa_id": "uvo__-5O_xxIpooE__camera_translation_direction__00", "video_id": "uvo__-5O_xxIpooE", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "up", "down", "left"], "frame_paths": ["media/17f5d1a1b68e377d5ed1/F0.jpg", "media/17f5d1a1b68e377d5ed1/F1.jpg", "media/17f5d1a1b68e377d5ed1/F2.jpg", "media/17f5d1a1b68e377d5ed1/F3.jpg", "media/17f5d1a1b68e377d5ed1/F4.jpg", "media/17f5d1a1b68e377d5ed1/F5.jpg", "media/17f5d1a1b68e377d5ed1/F6.jpg", "media/17f5d1a1b68e377d5ed1/F7.jpg", "media/17f5d1a1b68e377d5ed1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73bc90e72a291f15b3a4ce9e532ab91644e01119819ca89179bb19e62efe6b29"}
{"qa_id": "uvo__-5O_xxIpooE__camera_view_direction_change__01", "video_id": "uvo__-5O_xxIpooE", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/17f5d1a1b68e377d5ed1/F0.jpg", "media/17f5d1a1b68e377d5ed1/F1.jpg", "media/17f5d1a1b68e377d5ed1/F2.jpg", "media/17f5d1a1b68e377d5ed1/F3.jpg", "media/17f5d1a1b68e377d5ed1/F4.jpg", "media/17f5d1a1b68e377d5ed1/F5.jpg", "media/17f5d1a1b68e377d5ed1/F6.jpg", "media/17f5d1a1b68e377d5ed1/F7.jpg", "media/17f5d1a1b68e377d5ed1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "21b0eed225d3020fc4eb6acb972c4a23bd64d422a544b06ac2c41858aa50ecb3"}
{"qa_id": "uvo__-5O_xxIpooE__object_world_motion_direction__02", "video_id": "uvo__-5O_xxIpooE", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (person wearing an orange shirt and a black adidas cap) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "backward", "forward", "right"], "frame_paths": ["media/70ef3a52aa31c1d648fe/F0.jpg", "media/70ef3a52aa31c1d648fe/F1.jpg", "media/70ef3a52aa31c1d648fe/F2.jpg", "media/70ef3a52aa31c1d648fe/F3.jpg", "media/70ef3a52aa31c1d648fe/F4.jpg", "media/70ef3a52aa31c1d648fe/F5.jpg", "media/70ef3a52aa31c1d648fe/F6.jpg", "media/70ef3a52aa31c1d648fe/F7.jpg", "media/70ef3a52aa31c1d648fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (person wearing an orange shirt and a black adidas cap) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d64c68ef8a1575e27433c894e077b6a907b9dd5cc4aeb11b288332f376fde3e9"}
{"qa_id": "uvo__-5Tu_wiNnmg__camera_object_distance_change__00", "video_id": "uvo__-5Tu_wiNnmg", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (man in kilt and black jacket playing bagpipes) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/72d6292bfd5081eecdc9/F0.jpg", "media/72d6292bfd5081eecdc9/F1.jpg", "media/72d6292bfd5081eecdc9/F2.jpg", "media/72d6292bfd5081eecdc9/F3.jpg", "media/72d6292bfd5081eecdc9/F4.jpg", "media/72d6292bfd5081eecdc9/F5.jpg", "media/72d6292bfd5081eecdc9/F6.jpg", "media/72d6292bfd5081eecdc9/F7.jpg", "media/72d6292bfd5081eecdc9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (man in kilt and black jacket playing bagpipes) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "21fcbc609239483ca355f3b001e9927ede98a7a6ad4b86bb20f3e3bf2e7159cb"}
{"qa_id": "uvo__-5Tu_wiNnmg__camera_object_distance_change__01", "video_id": "uvo__-5Tu_wiNnmg", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object B (man in kilt and black jacket playing bagpipes) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f8ef7be79788e0c41b41/F0.jpg", "media/f8ef7be79788e0c41b41/F1.jpg", "media/f8ef7be79788e0c41b41/F2.jpg", "media/f8ef7be79788e0c41b41/F3.jpg", "media/f8ef7be79788e0c41b41/F4.jpg", "media/f8ef7be79788e0c41b41/F5.jpg", "media/f8ef7be79788e0c41b41/F6.jpg", "media/f8ef7be79788e0c41b41/F7.jpg", "media/f8ef7be79788e0c41b41/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5", "B": "g6"}, "requested_object_ids": ["g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (man in kilt and black jacket playing bagpipes) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1eb7f4142b825fff7a582628b19b0b6726e46d886123cca8b8e208edf1be0d8e"}
{"qa_id": "uvo__-5kButbj9OE__camera_translation_direction__00", "video_id": "uvo__-5kButbj9OE", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "up", "backward", "forward"], "frame_paths": ["media/ccc8d0345f4c8db30683/F0.jpg", "media/ccc8d0345f4c8db30683/F1.jpg", "media/ccc8d0345f4c8db30683/F2.jpg", "media/ccc8d0345f4c8db30683/F3.jpg", "media/ccc8d0345f4c8db30683/F4.jpg", "media/ccc8d0345f4c8db30683/F5.jpg", "media/ccc8d0345f4c8db30683/F6.jpg", "media/ccc8d0345f4c8db30683/F7.jpg", "media/ccc8d0345f4c8db30683/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "accf3e3582311cd41bd586a5749e43b7a501c1353d40f99b32f212d6e1828aae"}
{"qa_id": "uvo__-5kButbj9OE__camera_view_direction_change__01", "video_id": "uvo__-5kButbj9OE", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/ccc8d0345f4c8db30683/F0.jpg", "media/ccc8d0345f4c8db30683/F1.jpg", "media/ccc8d0345f4c8db30683/F2.jpg", "media/ccc8d0345f4c8db30683/F3.jpg", "media/ccc8d0345f4c8db30683/F4.jpg", "media/ccc8d0345f4c8db30683/F5.jpg", "media/ccc8d0345f4c8db30683/F6.jpg", "media/ccc8d0345f4c8db30683/F7.jpg", "media/ccc8d0345f4c8db30683/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c6c89cdaa9454d80370f8b0304157b478d24917d2ca062a4448248cfc708625e"}
{"qa_id": "uvo__-6VzhYhWDMY__camera_translation_direction__00", "video_id": "uvo__-6VzhYhWDMY", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "right", "up", "forward"], "frame_paths": ["media/03ef8bddd4ed1f90b306/F0.jpg", "media/03ef8bddd4ed1f90b306/F1.jpg", "media/03ef8bddd4ed1f90b306/F2.jpg", "media/03ef8bddd4ed1f90b306/F3.jpg", "media/03ef8bddd4ed1f90b306/F4.jpg", "media/03ef8bddd4ed1f90b306/F5.jpg", "media/03ef8bddd4ed1f90b306/F6.jpg", "media/03ef8bddd4ed1f90b306/F7.jpg", "media/03ef8bddd4ed1f90b306/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91ba34b6b6607a9201d7ba8083ca4e071b75ebcece5b44b60dd2bf634a8e9295"}
{"qa_id": "uvo__-6VzhYhWDMY__object_world_motion_direction__01", "video_id": "uvo__-6VzhYhWDMY", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in pink outfit) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "forward", "up", "left"], "frame_paths": ["media/60ec792676f60a18eb33/F0.jpg", "media/60ec792676f60a18eb33/F1.jpg", "media/60ec792676f60a18eb33/F2.jpg", "media/60ec792676f60a18eb33/F3.jpg", "media/60ec792676f60a18eb33/F4.jpg", "media/60ec792676f60a18eb33/F5.jpg", "media/60ec792676f60a18eb33/F6.jpg", "media/60ec792676f60a18eb33/F7.jpg", "media/60ec792676f60a18eb33/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in pink outfit) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5b5ca59a9d7c86d8b8348f423948984b207b6b5f61d929f37aa78109126eade"}
{"qa_id": "uvo__-6d7xfYKptE__camera_translation_direction__00", "video_id": "uvo__-6d7xfYKptE", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/b7a8b7ba913b6af6739d/F0.jpg", "media/b7a8b7ba913b6af6739d/F1.jpg", "media/b7a8b7ba913b6af6739d/F2.jpg", "media/b7a8b7ba913b6af6739d/F3.jpg", "media/b7a8b7ba913b6af6739d/F4.jpg", "media/b7a8b7ba913b6af6739d/F5.jpg", "media/b7a8b7ba913b6af6739d/F6.jpg", "media/b7a8b7ba913b6af6739d/F7.jpg", "media/b7a8b7ba913b6af6739d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d42062fc44a18f795eb6a8c383d7c94789a3fb7b2f1c46ac17de04038da08579"}
{"qa_id": "uvo__-7yP-JTSHj0__camera_view_direction_change__00", "video_id": "uvo__-7yP-JTSHj0", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/9c92390ebc2f539eeb46/F0.jpg", "media/9c92390ebc2f539eeb46/F1.jpg", "media/9c92390ebc2f539eeb46/F2.jpg", "media/9c92390ebc2f539eeb46/F3.jpg", "media/9c92390ebc2f539eeb46/F4.jpg", "media/9c92390ebc2f539eeb46/F5.jpg", "media/9c92390ebc2f539eeb46/F6.jpg", "media/9c92390ebc2f539eeb46/F7.jpg", "media/9c92390ebc2f539eeb46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "043300a1a66fbe16da2b9aacdd99d3d75b704882790f77cda9a7b3b2af548c85"}
{"qa_id": "uvo__-8n8BguY7_Y__camera_translation_direction__00", "video_id": "uvo__-8n8BguY7_Y", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/85e1254627b3b1cc0819/F0.jpg", "media/85e1254627b3b1cc0819/F1.jpg", "media/85e1254627b3b1cc0819/F2.jpg", "media/85e1254627b3b1cc0819/F3.jpg", "media/85e1254627b3b1cc0819/F4.jpg", "media/85e1254627b3b1cc0819/F5.jpg", "media/85e1254627b3b1cc0819/F6.jpg", "media/85e1254627b3b1cc0819/F7.jpg", "media/85e1254627b3b1cc0819/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4dd5750d2b1498617a0e867d02ecc0097ca868ab28f49107d979e34fc39f8433"}
{"qa_id": "uvo__-9DSHbAQ3F4__camera_translation_direction__00", "video_id": "uvo__-9DSHbAQ3F4", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "down", "backward", "right", "left"], "frame_paths": ["media/e966dd6ceafb3f4d0d49/F0.jpg", "media/e966dd6ceafb3f4d0d49/F1.jpg", "media/e966dd6ceafb3f4d0d49/F2.jpg", "media/e966dd6ceafb3f4d0d49/F3.jpg", "media/e966dd6ceafb3f4d0d49/F4.jpg", "media/e966dd6ceafb3f4d0d49/F5.jpg", "media/e966dd6ceafb3f4d0d49/F6.jpg", "media/e966dd6ceafb3f4d0d49/F7.jpg", "media/e966dd6ceafb3f4d0d49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. down\nD. backward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "14f7689688c7863362d20bb886c135b9c2a27eaea2e915bcdc12c7708bcff5b5"}
{"qa_id": "uvo__-9DSHbAQ3F4__camera_view_direction_change__01", "video_id": "uvo__-9DSHbAQ3F4", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/e966dd6ceafb3f4d0d49/F0.jpg", "media/e966dd6ceafb3f4d0d49/F1.jpg", "media/e966dd6ceafb3f4d0d49/F2.jpg", "media/e966dd6ceafb3f4d0d49/F3.jpg", "media/e966dd6ceafb3f4d0d49/F4.jpg", "media/e966dd6ceafb3f4d0d49/F5.jpg", "media/e966dd6ceafb3f4d0d49/F6.jpg", "media/e966dd6ceafb3f4d0d49/F7.jpg", "media/e966dd6ceafb3f4d0d49/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fec7177b28cd260bac8abbd97c394e95ca01afdba0cbfecdcc2d079d3b0a8ef6"}
{"qa_id": "uvo__-A-5wR2-u5M__camera_view_direction_change__00", "video_id": "uvo__-A-5wR2-u5M", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/ef37bb73186efbb3b3a2/F0.jpg", "media/ef37bb73186efbb3b3a2/F1.jpg", "media/ef37bb73186efbb3b3a2/F2.jpg", "media/ef37bb73186efbb3b3a2/F3.jpg", "media/ef37bb73186efbb3b3a2/F4.jpg", "media/ef37bb73186efbb3b3a2/F5.jpg", "media/ef37bb73186efbb3b3a2/F6.jpg", "media/ef37bb73186efbb3b3a2/F7.jpg", "media/ef37bb73186efbb3b3a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a54a241ed1bb9fdb26ed772ff8ea73a8d73d4bbd30616749849263a06af7ea21"}
{"qa_id": "uvo__-BkcYlaP_gI__camera_translation_direction__00", "video_id": "uvo__-BkcYlaP_gI", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "forward", "up"], "frame_paths": ["media/9a1333ca38b075329cdd/F0.jpg", "media/9a1333ca38b075329cdd/F1.jpg", "media/9a1333ca38b075329cdd/F2.jpg", "media/9a1333ca38b075329cdd/F3.jpg", "media/9a1333ca38b075329cdd/F4.jpg", "media/9a1333ca38b075329cdd/F5.jpg", "media/9a1333ca38b075329cdd/F6.jpg", "media/9a1333ca38b075329cdd/F7.jpg", "media/9a1333ca38b075329cdd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "558d2e9ae9aaccb78e370d716bfc978b03bf147a39e924ce80677ac24d7bb150"}
{"qa_id": "uvo__-BkcYlaP_gI__camera_object_distance_change__02", "video_id": "uvo__-BkcYlaP_gI", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (person in a blue jacket with white stripes standing on a blue sled) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/721a9094ad46965365be/F0.jpg", "media/721a9094ad46965365be/F1.jpg", "media/721a9094ad46965365be/F2.jpg", "media/721a9094ad46965365be/F3.jpg", "media/721a9094ad46965365be/F4.jpg", "media/721a9094ad46965365be/F5.jpg", "media/721a9094ad46965365be/F6.jpg", "media/721a9094ad46965365be/F7.jpg", "media/721a9094ad46965365be/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (person in a blue jacket with white stripes standing on a blue sled) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4735833a631c68b65189d43ceb602124e48daaabb8fc6f61a279f352e045d93a"}
{"qa_id": "uvo__-Cw8Z-5ShUM__camera_view_direction_change__00", "video_id": "uvo__-Cw8Z-5ShUM", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/019580a20ffae915987c/F0.jpg", "media/019580a20ffae915987c/F1.jpg", "media/019580a20ffae915987c/F2.jpg", "media/019580a20ffae915987c/F3.jpg", "media/019580a20ffae915987c/F4.jpg", "media/019580a20ffae915987c/F5.jpg", "media/019580a20ffae915987c/F6.jpg", "media/019580a20ffae915987c/F7.jpg", "media/019580a20ffae915987c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "uvo__-Eh3Bruf27Y__camera_view_direction_change__00", "video_id": "uvo__-Eh3Bruf27Y", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/ea8cf0f200b1d06ce947/F0.jpg", "media/ea8cf0f200b1d06ce947/F1.jpg", "media/ea8cf0f200b1d06ce947/F2.jpg", "media/ea8cf0f200b1d06ce947/F3.jpg", "media/ea8cf0f200b1d06ce947/F4.jpg", "media/ea8cf0f200b1d06ce947/F5.jpg", "media/ea8cf0f200b1d06ce947/F6.jpg", "media/ea8cf0f200b1d06ce947/F7.jpg", "media/ea8cf0f200b1d06ce947/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16a4ca6eeb0d84a0d96a1b095761436c4777b97628640bbab69e354a88eb5d55"}
{"qa_id": "uvo__-Eh3Bruf27Y__object_world_motion_direction__01", "video_id": "uvo__-Eh3Bruf27Y", "task": "object_world_motion_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object A (teal excavator arm) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "forward", "down"], "frame_paths": ["media/d9c57ec4163a7615af9e/F0.jpg", "media/d9c57ec4163a7615af9e/F1.jpg", "media/d9c57ec4163a7615af9e/F2.jpg", "media/d9c57ec4163a7615af9e/F3.jpg", "media/d9c57ec4163a7615af9e/F4.jpg", "media/d9c57ec4163a7615af9e/F5.jpg", "media/d9c57ec4163a7615af9e/F6.jpg", "media/d9c57ec4163a7615af9e/F7.jpg", "media/d9c57ec4163a7615af9e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object A (teal excavator arm) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b9ea5483a2ad6c67cba9b369da13150abdb5dcb12c90a5e4383892f32fb2fa1f"}
{"qa_id": "uvo__-F0Hh-YYbN4__camera_view_direction_change__00", "video_id": "uvo__-F0Hh-YYbN4", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/8a8cbc2d5c6ed75fd82b/F0.jpg", "media/8a8cbc2d5c6ed75fd82b/F1.jpg", "media/8a8cbc2d5c6ed75fd82b/F2.jpg", "media/8a8cbc2d5c6ed75fd82b/F3.jpg", "media/8a8cbc2d5c6ed75fd82b/F4.jpg", "media/8a8cbc2d5c6ed75fd82b/F5.jpg", "media/8a8cbc2d5c6ed75fd82b/F6.jpg", "media/8a8cbc2d5c6ed75fd82b/F7.jpg", "media/8a8cbc2d5c6ed75fd82b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5217b9ae1d03633501a940bc1f7b52d23829ee042f72ee0a3ba9b28bdb58d7c6"}
{"qa_id": "uvo__-Fp5reKg40w__camera_translation_direction__00", "video_id": "uvo__-Fp5reKg40w", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "forward", "backward"], "frame_paths": ["media/103e8338150b8be951e9/F0.jpg", "media/103e8338150b8be951e9/F1.jpg", "media/103e8338150b8be951e9/F2.jpg", "media/103e8338150b8be951e9/F3.jpg", "media/103e8338150b8be951e9/F4.jpg", "media/103e8338150b8be951e9/F5.jpg", "media/103e8338150b8be951e9/F6.jpg", "media/103e8338150b8be951e9/F7.jpg", "media/103e8338150b8be951e9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "47cf371585fc218552e20c2feab9b14d0e4a1bec166d7f8254c3e65ece94570f"}
{"qa_id": "uvo__-GL4Rp-F50k__object_pair_distance_change__00", "video_id": "uvo__-GL4Rp-F50k", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (person in black outfit) and object B (person in dark green jacket) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2bb6caf0c7a8c1aeaf37/F0.jpg", "media/2bb6caf0c7a8c1aeaf37/F1.jpg", "media/2bb6caf0c7a8c1aeaf37/F2.jpg", "media/2bb6caf0c7a8c1aeaf37/F3.jpg", "media/2bb6caf0c7a8c1aeaf37/F4.jpg", "media/2bb6caf0c7a8c1aeaf37/F5.jpg", "media/2bb6caf0c7a8c1aeaf37/F6.jpg", "media/2bb6caf0c7a8c1aeaf37/F7.jpg", "media/2bb6caf0c7a8c1aeaf37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (person in black outfit) and object B (person in dark green jacket) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3ffcd555d06e5baab694001f21993a4849dc1837a024181d95f3a06d8c2a005"}
{"qa_id": "uvo__-GL4Rp-F50k__object_travel_distance_comparison__01", "video_id": "uvo__-GL4Rp-F50k", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black outfit) or B (person in dark green jacket)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2bb6caf0c7a8c1aeaf37/F0.jpg", "media/2bb6caf0c7a8c1aeaf37/F1.jpg", "media/2bb6caf0c7a8c1aeaf37/F2.jpg", "media/2bb6caf0c7a8c1aeaf37/F3.jpg", "media/2bb6caf0c7a8c1aeaf37/F4.jpg", "media/2bb6caf0c7a8c1aeaf37/F5.jpg", "media/2bb6caf0c7a8c1aeaf37/F6.jpg", "media/2bb6caf0c7a8c1aeaf37/F7.jpg", "media/2bb6caf0c7a8c1aeaf37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black outfit) or B (person in dark green jacket)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9dea6cfb183a4f0145cb5437b29054d0eca5cd04c270f4d6e83cdc3ca1e5780"}
{"qa_id": "uvo__-IrgG_SljXU__camera_translation_direction__00", "video_id": "uvo__-IrgG_SljXU", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "right", "backward", "down", "left"], "frame_paths": ["media/8b1fd4ec43c4cf4b2a19/F0.jpg", "media/8b1fd4ec43c4cf4b2a19/F1.jpg", "media/8b1fd4ec43c4cf4b2a19/F2.jpg", "media/8b1fd4ec43c4cf4b2a19/F3.jpg", "media/8b1fd4ec43c4cf4b2a19/F4.jpg", "media/8b1fd4ec43c4cf4b2a19/F5.jpg", "media/8b1fd4ec43c4cf4b2a19/F6.jpg", "media/8b1fd4ec43c4cf4b2a19/F7.jpg", "media/8b1fd4ec43c4cf4b2a19/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. right\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4663d03a83192a6c6605ea1da45646a036adcfbf288456a1dc5e546549897189"}
{"qa_id": "uvo__-JfMuIHK7Yg__camera_translation_direction__00", "video_id": "uvo__-JfMuIHK7Yg", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "up", "left", "forward"], "frame_paths": ["media/74f7ed89a103225d9f3e/F0.jpg", "media/74f7ed89a103225d9f3e/F1.jpg", "media/74f7ed89a103225d9f3e/F2.jpg", "media/74f7ed89a103225d9f3e/F3.jpg", "media/74f7ed89a103225d9f3e/F4.jpg", "media/74f7ed89a103225d9f3e/F5.jpg", "media/74f7ed89a103225d9f3e/F6.jpg", "media/74f7ed89a103225d9f3e/F7.jpg", "media/74f7ed89a103225d9f3e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ac18d43b1ecee4a5d3aafdfc0fce7b4f5a0ccbf0563f444cc1d18e34035d7954"}
{"qa_id": "uvo__-JfMuIHK7Yg__camera_view_direction_change__01", "video_id": "uvo__-JfMuIHK7Yg", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/74f7ed89a103225d9f3e/F0.jpg", "media/74f7ed89a103225d9f3e/F1.jpg", "media/74f7ed89a103225d9f3e/F2.jpg", "media/74f7ed89a103225d9f3e/F3.jpg", "media/74f7ed89a103225d9f3e/F4.jpg", "media/74f7ed89a103225d9f3e/F5.jpg", "media/74f7ed89a103225d9f3e/F6.jpg", "media/74f7ed89a103225d9f3e/F7.jpg", "media/74f7ed89a103225d9f3e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fca55b1611ae8dd2346bf922298894d229a7f1fe48f897ccdbc4dcbbbea48d9"}
{"qa_id": "uvo__-L05Cc8huXg__object_world_motion_direction__00", "video_id": "uvo__-L05Cc8huXg", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (musher in purple jacket and orange hat) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "up", "down"], "frame_paths": ["media/98a30e345c6385dbf6c3/F0.jpg", "media/98a30e345c6385dbf6c3/F1.jpg", "media/98a30e345c6385dbf6c3/F2.jpg", "media/98a30e345c6385dbf6c3/F3.jpg", "media/98a30e345c6385dbf6c3/F4.jpg", "media/98a30e345c6385dbf6c3/F5.jpg", "media/98a30e345c6385dbf6c3/F6.jpg", "media/98a30e345c6385dbf6c3/F7.jpg", "media/98a30e345c6385dbf6c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (musher in purple jacket and orange hat) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b1344a54afa283f0bf9c165eaaee0960f969b2aad121b934faa6aee868d54cd1"}
{"qa_id": "uvo__-LenAZs2OFM__camera_view_direction_change__00", "video_id": "uvo__-LenAZs2OFM", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/47ec54f04479a723bccd/F0.jpg", "media/47ec54f04479a723bccd/F1.jpg", "media/47ec54f04479a723bccd/F2.jpg", "media/47ec54f04479a723bccd/F3.jpg", "media/47ec54f04479a723bccd/F4.jpg", "media/47ec54f04479a723bccd/F5.jpg", "media/47ec54f04479a723bccd/F6.jpg", "media/47ec54f04479a723bccd/F7.jpg", "media/47ec54f04479a723bccd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b82c7450be38b833c011d21c2de1a6440bfbeb318b187e60065e1c82616d86a9"}
{"qa_id": "uvo__-MYJtLnxUqY__camera_view_direction_change__00", "video_id": "uvo__-MYJtLnxUqY", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/1dc34e79da6d8113fcab/F0.jpg", "media/1dc34e79da6d8113fcab/F1.jpg", "media/1dc34e79da6d8113fcab/F2.jpg", "media/1dc34e79da6d8113fcab/F3.jpg", "media/1dc34e79da6d8113fcab/F4.jpg", "media/1dc34e79da6d8113fcab/F5.jpg", "media/1dc34e79da6d8113fcab/F6.jpg", "media/1dc34e79da6d8113fcab/F7.jpg", "media/1dc34e79da6d8113fcab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "uvo__-MYJtLnxUqY__camera_object_distance_change__01", "video_id": "uvo__-MYJtLnxUqY", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object A (child in pink and grey raglan shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/85f511cddf7ef6ceb950/F0.jpg", "media/85f511cddf7ef6ceb950/F1.jpg", "media/85f511cddf7ef6ceb950/F2.jpg", "media/85f511cddf7ef6ceb950/F3.jpg", "media/85f511cddf7ef6ceb950/F4.jpg", "media/85f511cddf7ef6ceb950/F5.jpg", "media/85f511cddf7ef6ceb950/F6.jpg", "media/85f511cddf7ef6ceb950/F7.jpg", "media/85f511cddf7ef6ceb950/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (child in pink and grey raglan shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2853ab0001e1bdc797541ef21ad665a1f4c241eed99d99504343b7b96c7cc24e"}
{"qa_id": "uvo__-MYJtLnxUqY__camera_object_distance_change__02", "video_id": "uvo__-MYJtLnxUqY", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object B (person in pink patterned shirt with ponytail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/2c74df46178492123a28/F0.jpg", "media/2c74df46178492123a28/F1.jpg", "media/2c74df46178492123a28/F2.jpg", "media/2c74df46178492123a28/F3.jpg", "media/2c74df46178492123a28/F4.jpg", "media/2c74df46178492123a28/F5.jpg", "media/2c74df46178492123a28/F6.jpg", "media/2c74df46178492123a28/F7.jpg", "media/2c74df46178492123a28/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object B (person in pink patterned shirt with ponytail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "57271c26abaafc5910e9e724db64379ea286e7c77c1249f0722a0304f28e9ae4"}
{"qa_id": "uvo__-MYJtLnxUqY__object_pair_distance_change__03", "video_id": "uvo__-MYJtLnxUqY", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did the 3D distance between object A (child in pink and grey raglan shirt) and object B (person in pink patterned shirt with ponytail) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/b0cf6345683e580fe661/F0.jpg", "media/b0cf6345683e580fe661/F1.jpg", "media/b0cf6345683e580fe661/F2.jpg", "media/b0cf6345683e580fe661/F3.jpg", "media/b0cf6345683e580fe661/F4.jpg", "media/b0cf6345683e580fe661/F5.jpg", "media/b0cf6345683e580fe661/F6.jpg", "media/b0cf6345683e580fe661/F7.jpg", "media/b0cf6345683e580fe661/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (child in pink and grey raglan shirt) and object B (person in pink patterned shirt with ponytail) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0f671d609ac2670042aa9b4511b191bb90847022ab2dde34f9c2dad97be64334"}
{"qa_id": "uvo__-MYJtLnxUqY__object_travel_distance_comparison__04", "video_id": "uvo__-MYJtLnxUqY", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in pink and grey raglan shirt) or B (person in pink patterned shirt with ponytail)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/b0cf6345683e580fe661/F0.jpg", "media/b0cf6345683e580fe661/F1.jpg", "media/b0cf6345683e580fe661/F2.jpg", "media/b0cf6345683e580fe661/F3.jpg", "media/b0cf6345683e580fe661/F4.jpg", "media/b0cf6345683e580fe661/F5.jpg", "media/b0cf6345683e580fe661/F6.jpg", "media/b0cf6345683e580fe661/F7.jpg", "media/b0cf6345683e580fe661/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in pink and grey raglan shirt) or B (person in pink patterned shirt with ponytail)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd7f08db9654c8efb6cd8b745b50235bdad37ddf4e9448fcd0b2ab1d1ab6f0a1"}
{"qa_id": "uvo__-MomTQ4exNk__camera_translation_direction__00", "video_id": "uvo__-MomTQ4exNk", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "backward", "left", "down"], "frame_paths": ["media/c10ce212e78ff61229c0/F0.jpg", "media/c10ce212e78ff61229c0/F1.jpg", "media/c10ce212e78ff61229c0/F2.jpg", "media/c10ce212e78ff61229c0/F3.jpg", "media/c10ce212e78ff61229c0/F4.jpg", "media/c10ce212e78ff61229c0/F5.jpg", "media/c10ce212e78ff61229c0/F6.jpg", "media/c10ce212e78ff61229c0/F7.jpg", "media/c10ce212e78ff61229c0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "28213c8ce9ef9aa3fe0c84b9b3f96546b613cea77313aa5c6fd649abe6b33407"}
{"qa_id": "uvo__-PJZHqWvhdc__object_world_motion_direction__00", "video_id": "uvo__-PJZHqWvhdc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (ice hockey player in grey jersey and white helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "down", "backward", "forward"], "frame_paths": ["media/d579978369d015fa5208/F0.jpg", "media/d579978369d015fa5208/F1.jpg", "media/d579978369d015fa5208/F2.jpg", "media/d579978369d015fa5208/F3.jpg", "media/d579978369d015fa5208/F4.jpg", "media/d579978369d015fa5208/F5.jpg", "media/d579978369d015fa5208/F6.jpg", "media/d579978369d015fa5208/F7.jpg", "media/d579978369d015fa5208/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (ice hockey player in grey jersey and white helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4eb03d2755318283616c58fe6fc26b5802d91040b70e78791ff74a51da97c5ca"}
{"qa_id": "uvo__-PJZHqWvhdc__camera_object_distance_change__01", "video_id": "uvo__-PJZHqWvhdc", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object A (ice hockey player in grey jersey and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d579978369d015fa5208/F0.jpg", "media/d579978369d015fa5208/F1.jpg", "media/d579978369d015fa5208/F2.jpg", "media/d579978369d015fa5208/F3.jpg", "media/d579978369d015fa5208/F4.jpg", "media/d579978369d015fa5208/F5.jpg", "media/d579978369d015fa5208/F6.jpg", "media/d579978369d015fa5208/F7.jpg", "media/d579978369d015fa5208/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (ice hockey player in grey jersey and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "906215d9f998d76f16b529015bc1bc8d00511641fa2686b175ddcd15a39e0503"}
{"qa_id": "uvo__-QKOeEPeNa8__camera_view_direction_change__00", "video_id": "uvo__-QKOeEPeNa8", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/11fea3cf83762fee98be/F0.jpg", "media/11fea3cf83762fee98be/F1.jpg", "media/11fea3cf83762fee98be/F2.jpg", "media/11fea3cf83762fee98be/F3.jpg", "media/11fea3cf83762fee98be/F4.jpg", "media/11fea3cf83762fee98be/F5.jpg", "media/11fea3cf83762fee98be/F6.jpg", "media/11fea3cf83762fee98be/F7.jpg", "media/11fea3cf83762fee98be/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f50dcc4587b168ab9459635029bd7a6e5fa68bafe0ef7289da83ccd89fc17f1f"}
{"qa_id": "uvo__-RhNquKdM0M__camera_view_direction_change__00", "video_id": "uvo__-RhNquKdM0M", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/442d487c8bd807aa6077/F0.jpg", "media/442d487c8bd807aa6077/F1.jpg", "media/442d487c8bd807aa6077/F2.jpg", "media/442d487c8bd807aa6077/F3.jpg", "media/442d487c8bd807aa6077/F4.jpg", "media/442d487c8bd807aa6077/F5.jpg", "media/442d487c8bd807aa6077/F6.jpg", "media/442d487c8bd807aa6077/F7.jpg", "media/442d487c8bd807aa6077/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fec7177b28cd260bac8abbd97c394e95ca01afdba0cbfecdcc2d079d3b0a8ef6"}
{"qa_id": "uvo__-SGc-osF_VM__camera_object_distance_change__00", "video_id": "uvo__-SGc-osF_VM", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did object B (person in yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8a44c6d231588bd5d906/F0.jpg", "media/8a44c6d231588bd5d906/F1.jpg", "media/8a44c6d231588bd5d906/F2.jpg", "media/8a44c6d231588bd5d906/F3.jpg", "media/8a44c6d231588bd5d906/F4.jpg", "media/8a44c6d231588bd5d906/F5.jpg", "media/8a44c6d231588bd5d906/F6.jpg", "media/8a44c6d231588bd5d906/F7.jpg", "media/8a44c6d231588bd5d906/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object B (person in yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "823b981896dfe32c3bd5019725fe642350ee3dc024eee6921785087998b82ad9"}
{"qa_id": "uvo__-SWG69JZ_vk__camera_view_direction_change__00", "video_id": "uvo__-SWG69JZ_vk", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/49281f25ed69b53f8107/F0.jpg", "media/49281f25ed69b53f8107/F1.jpg", "media/49281f25ed69b53f8107/F2.jpg", "media/49281f25ed69b53f8107/F3.jpg", "media/49281f25ed69b53f8107/F4.jpg", "media/49281f25ed69b53f8107/F5.jpg", "media/49281f25ed69b53f8107/F6.jpg", "media/49281f25ed69b53f8107/F7.jpg", "media/49281f25ed69b53f8107/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fca55b1611ae8dd2346bf922298894d229a7f1fe48f897ccdbc4dcbbbea48d9"}
{"qa_id": "uvo__-T5iLC1Bm0g__camera_translation_direction__00", "video_id": "uvo__-T5iLC1Bm0g", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "forward", "left", "down"], "frame_paths": ["media/842828935a9cd5d8b17c/F0.jpg", "media/842828935a9cd5d8b17c/F1.jpg", "media/842828935a9cd5d8b17c/F2.jpg", "media/842828935a9cd5d8b17c/F3.jpg", "media/842828935a9cd5d8b17c/F4.jpg", "media/842828935a9cd5d8b17c/F5.jpg", "media/842828935a9cd5d8b17c/F6.jpg", "media/842828935a9cd5d8b17c/F7.jpg", "media/842828935a9cd5d8b17c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ad652acbe4a2b89721fdf30474cb4ac7cc386f4e628546cd8b0507aa30602b70"}
{"qa_id": "uvo__-T5iLC1Bm0g__camera_view_direction_change__01", "video_id": "uvo__-T5iLC1Bm0g", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/842828935a9cd5d8b17c/F0.jpg", "media/842828935a9cd5d8b17c/F1.jpg", "media/842828935a9cd5d8b17c/F2.jpg", "media/842828935a9cd5d8b17c/F3.jpg", "media/842828935a9cd5d8b17c/F4.jpg", "media/842828935a9cd5d8b17c/F5.jpg", "media/842828935a9cd5d8b17c/F6.jpg", "media/842828935a9cd5d8b17c/F7.jpg", "media/842828935a9cd5d8b17c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64f88ed5d07ee48af2c22e988f262f3514b3c8739559f9d0fe2ae33efb227b09"}
{"qa_id": "uvo__-T5iLC1Bm0g__camera_object_distance_change__02", "video_id": "uvo__-T5iLC1Bm0g", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, did object A (person in red and white striped shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/73181b000a25ebda0645/F0.jpg", "media/73181b000a25ebda0645/F1.jpg", "media/73181b000a25ebda0645/F2.jpg", "media/73181b000a25ebda0645/F3.jpg", "media/73181b000a25ebda0645/F4.jpg", "media/73181b000a25ebda0645/F5.jpg", "media/73181b000a25ebda0645/F6.jpg", "media/73181b000a25ebda0645/F7.jpg", "media/73181b000a25ebda0645/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, did object A (person in red and white striped shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dbd29558e4a553df429702b4d96cb646386d1b5cab0151fd1a984219326c7fd6"}
{"qa_id": "uvo__-TUCIO1M01o__camera_translation_direction__00", "video_id": "uvo__-TUCIO1M01o", "task": "camera_translation_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "down", "up", "forward"], "frame_paths": ["media/c1e7246510326117e289/F0.jpg", "media/c1e7246510326117e289/F1.jpg", "media/c1e7246510326117e289/F2.jpg", "media/c1e7246510326117e289/F3.jpg", "media/c1e7246510326117e289/F4.jpg", "media/c1e7246510326117e289/F5.jpg", "media/c1e7246510326117e289/F6.jpg", "media/c1e7246510326117e289/F7.jpg", "media/c1e7246510326117e289/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "804e155baa9e6a4c0d6f22801f04e7a2cbb53dbadff87fc27bc1a96ab2608f78"}
{"qa_id": "uvo__-TUCIO1M01o__camera_view_direction_change__01", "video_id": "uvo__-TUCIO1M01o", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/c1e7246510326117e289/F0.jpg", "media/c1e7246510326117e289/F1.jpg", "media/c1e7246510326117e289/F2.jpg", "media/c1e7246510326117e289/F3.jpg", "media/c1e7246510326117e289/F4.jpg", "media/c1e7246510326117e289/F5.jpg", "media/c1e7246510326117e289/F6.jpg", "media/c1e7246510326117e289/F7.jpg", "media/c1e7246510326117e289/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0cec6af615458e9b9fa6cb3983791a9d3f458d2a3e519e16436ce5685bae426a"}
{"qa_id": "uvo__-TUCIO1M01o__object_world_motion_direction__02", "video_id": "uvo__-TUCIO1M01o", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in white long-sleeve shirt and black pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "forward", "down", "up"], "frame_paths": ["media/224c3a1ec229b848bdff/F0.jpg", "media/224c3a1ec229b848bdff/F1.jpg", "media/224c3a1ec229b848bdff/F2.jpg", "media/224c3a1ec229b848bdff/F3.jpg", "media/224c3a1ec229b848bdff/F4.jpg", "media/224c3a1ec229b848bdff/F5.jpg", "media/224c3a1ec229b848bdff/F6.jpg", "media/224c3a1ec229b848bdff/F7.jpg", "media/224c3a1ec229b848bdff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (person in white long-sleeve shirt and black pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "027fc1eed89c51415394570c5584298c7b373cfd77d1c03bf10193ffa4db5e64"}
{"qa_id": "uvo__-TdSou0hyEM__object_world_motion_direction__00", "video_id": "uvo__-TdSou0hyEM", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (small bicycle with orange wheels) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "up", "down", "left"], "frame_paths": ["media/20aed9fdf6d3708a9944/F0.jpg", "media/20aed9fdf6d3708a9944/F1.jpg", "media/20aed9fdf6d3708a9944/F2.jpg", "media/20aed9fdf6d3708a9944/F3.jpg", "media/20aed9fdf6d3708a9944/F4.jpg", "media/20aed9fdf6d3708a9944/F5.jpg", "media/20aed9fdf6d3708a9944/F6.jpg", "media/20aed9fdf6d3708a9944/F7.jpg", "media/20aed9fdf6d3708a9944/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (small bicycle with orange wheels) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f0594cc82ed6fc3b8afb1d178fd0bae6e8326f792e0a879abc466473b219110"}
{"qa_id": "uvo__-TdSou0hyEM__object_world_motion_direction__01", "video_id": "uvo__-TdSou0hyEM", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object B (child in blue jacket and green helmet riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "up", "forward", "down"], "frame_paths": ["media/4f33afef0ddfc3718c4e/F0.jpg", "media/4f33afef0ddfc3718c4e/F1.jpg", "media/4f33afef0ddfc3718c4e/F2.jpg", "media/4f33afef0ddfc3718c4e/F3.jpg", "media/4f33afef0ddfc3718c4e/F4.jpg", "media/4f33afef0ddfc3718c4e/F5.jpg", "media/4f33afef0ddfc3718c4e/F6.jpg", "media/4f33afef0ddfc3718c4e/F7.jpg", "media/4f33afef0ddfc3718c4e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object B (child in blue jacket and green helmet riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b2907e63f5e88bb22639c0941211272bd927d6ab5c1bfdd08b142dca397f63f5"}
{"qa_id": "uvo__-UVoHU_lcfQ__object_world_motion_direction__00", "video_id": "uvo__-UVoHU_lcfQ", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (sailboat with a white and blue sail) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "down", "up", "backward"], "frame_paths": ["media/1c98f7ca1a598cf8b943/F0.jpg", "media/1c98f7ca1a598cf8b943/F1.jpg", "media/1c98f7ca1a598cf8b943/F2.jpg", "media/1c98f7ca1a598cf8b943/F3.jpg", "media/1c98f7ca1a598cf8b943/F4.jpg", "media/1c98f7ca1a598cf8b943/F5.jpg", "media/1c98f7ca1a598cf8b943/F6.jpg", "media/1c98f7ca1a598cf8b943/F7.jpg", "media/1c98f7ca1a598cf8b943/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (sailboat with a white and blue sail) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "00ec68ed46034a562ccf30c38ffc83b707fa0bca2a624799e1d654dc6606ac0f"}
{"qa_id": "uvo__-UVoHU_lcfQ__camera_object_distance_change__01", "video_id": "uvo__-UVoHU_lcfQ", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, did object A (sailboat with a white and blue sail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/1c98f7ca1a598cf8b943/F0.jpg", "media/1c98f7ca1a598cf8b943/F1.jpg", "media/1c98f7ca1a598cf8b943/F2.jpg", "media/1c98f7ca1a598cf8b943/F3.jpg", "media/1c98f7ca1a598cf8b943/F4.jpg", "media/1c98f7ca1a598cf8b943/F5.jpg", "media/1c98f7ca1a598cf8b943/F6.jpg", "media/1c98f7ca1a598cf8b943/F7.jpg", "media/1c98f7ca1a598cf8b943/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (sailboat with a white and blue sail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "32e75730b5174707396c7f0e73d528d5f59873433666a20d58bb3f8ba0077d29"}
{"qa_id": "uvo__-V-mROa0TzQ__camera_view_direction_change__00", "video_id": "uvo__-V-mROa0TzQ", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/278d7cf7af08cccfb3eb/F0.jpg", "media/278d7cf7af08cccfb3eb/F1.jpg", "media/278d7cf7af08cccfb3eb/F2.jpg", "media/278d7cf7af08cccfb3eb/F3.jpg", "media/278d7cf7af08cccfb3eb/F4.jpg", "media/278d7cf7af08cccfb3eb/F5.jpg", "media/278d7cf7af08cccfb3eb/F6.jpg", "media/278d7cf7af08cccfb3eb/F7.jpg", "media/278d7cf7af08cccfb3eb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b753cba65d52068fe6729fa64acba3f9e6e5941e267c052876d88c95ba60a2f1"}
{"qa_id": "uvo__-V44aYCVuWU__camera_translation_direction__00", "video_id": "uvo__-V44aYCVuWU", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "forward", "down"], "frame_paths": ["media/5500db46905084c2d40d/F0.jpg", "media/5500db46905084c2d40d/F1.jpg", "media/5500db46905084c2d40d/F2.jpg", "media/5500db46905084c2d40d/F3.jpg", "media/5500db46905084c2d40d/F4.jpg", "media/5500db46905084c2d40d/F5.jpg", "media/5500db46905084c2d40d/F6.jpg", "media/5500db46905084c2d40d/F7.jpg", "media/5500db46905084c2d40d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5fba0a4a608c216327328badd02f077cfd357da575a714891867eaf2181c4238"}
{"qa_id": "uvo__-WkSAiei4r4__camera_translation_direction__00", "video_id": "uvo__-WkSAiei4r4", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "right", "up", "down", "backward"], "frame_paths": ["media/4854148e09ad4d845d04/F0.jpg", "media/4854148e09ad4d845d04/F1.jpg", "media/4854148e09ad4d845d04/F2.jpg", "media/4854148e09ad4d845d04/F3.jpg", "media/4854148e09ad4d845d04/F4.jpg", "media/4854148e09ad4d845d04/F5.jpg", "media/4854148e09ad4d845d04/F6.jpg", "media/4854148e09ad4d845d04/F7.jpg", "media/4854148e09ad4d845d04/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. right\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96521ce0d4bfbdc99b5b2bc494bc107f027b82764390acd5c6101fb4d85448b1"}
{"qa_id": "uvo__-XmBkxUhfMQ__camera_view_direction_change__00", "video_id": "uvo__-XmBkxUhfMQ", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/c7898e5082021defb679/F0.jpg", "media/c7898e5082021defb679/F1.jpg", "media/c7898e5082021defb679/F2.jpg", "media/c7898e5082021defb679/F3.jpg", "media/c7898e5082021defb679/F4.jpg", "media/c7898e5082021defb679/F5.jpg", "media/c7898e5082021defb679/F6.jpg", "media/c7898e5082021defb679/F7.jpg", "media/c7898e5082021defb679/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d03c1debb0028d8e206939967b08c855bc5443c5eb44121314c4a8896b32779"}
{"qa_id": "uvo__-YCuMrVTf0U__camera_translation_direction__00", "video_id": "uvo__-YCuMrVTf0U", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "forward", "right", "down"], "frame_paths": ["media/725b000afdd57274d8f0/F0.jpg", "media/725b000afdd57274d8f0/F1.jpg", "media/725b000afdd57274d8f0/F2.jpg", "media/725b000afdd57274d8f0/F3.jpg", "media/725b000afdd57274d8f0/F4.jpg", "media/725b000afdd57274d8f0/F5.jpg", "media/725b000afdd57274d8f0/F6.jpg", "media/725b000afdd57274d8f0/F7.jpg", "media/725b000afdd57274d8f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2ca87018415ac86a62c8f399f3c2c4d253fcabef8b217fd81ac6d5a51eef3345"}
{"qa_id": "uvo__-YGnDKPOEd4__object_world_motion_direction__00", "video_id": "uvo__-YGnDKPOEd4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in red shirt and black shorts playing tennis) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "down", "backward", "right", "up"], "frame_paths": ["media/4148e60ff21909e46b77/F0.jpg", "media/4148e60ff21909e46b77/F1.jpg", "media/4148e60ff21909e46b77/F2.jpg", "media/4148e60ff21909e46b77/F3.jpg", "media/4148e60ff21909e46b77/F4.jpg", "media/4148e60ff21909e46b77/F5.jpg", "media/4148e60ff21909e46b77/F6.jpg", "media/4148e60ff21909e46b77/F7.jpg", "media/4148e60ff21909e46b77/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in red shirt and black shorts playing tennis) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. down\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d36f4995534b51412926a3023ca6b4d4a4382eef07fe3f85cd1f07c184eb1705"}
{"qa_id": "uvo__-YtMJtz0C0Y__camera_view_direction_change__00", "video_id": "uvo__-YtMJtz0C0Y", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/6121bb5d894028633771/F0.jpg", "media/6121bb5d894028633771/F1.jpg", "media/6121bb5d894028633771/F2.jpg", "media/6121bb5d894028633771/F3.jpg", "media/6121bb5d894028633771/F4.jpg", "media/6121bb5d894028633771/F5.jpg", "media/6121bb5d894028633771/F6.jpg", "media/6121bb5d894028633771/F7.jpg", "media/6121bb5d894028633771/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "532dc7fc44fe3862cb953f79420b6cde00e93b6fc46897f8938fc3912831e120"}
{"qa_id": "uvo__-YtMJtz0C0Y__object_world_motion_direction__01", "video_id": "uvo__-YtMJtz0C0Y", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in dark clothing sledding down a snowy road) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "forward", "backward"], "frame_paths": ["media/40f93ef1789faf305ab5/F0.jpg", "media/40f93ef1789faf305ab5/F1.jpg", "media/40f93ef1789faf305ab5/F2.jpg", "media/40f93ef1789faf305ab5/F3.jpg", "media/40f93ef1789faf305ab5/F4.jpg", "media/40f93ef1789faf305ab5/F5.jpg", "media/40f93ef1789faf305ab5/F6.jpg", "media/40f93ef1789faf305ab5/F7.jpg", "media/40f93ef1789faf305ab5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in dark clothing sledding down a snowy road) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1aaf46592e1b8d2ece9b179a88ed96bc7b2a4f961280c2bd523ec57a244d6537"}
{"qa_id": "uvo__-YtMJtz0C0Y__camera_object_distance_change__02", "video_id": "uvo__-YtMJtz0C0Y", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F7, did object A (person in dark clothing sledding down a snowy road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/40f93ef1789faf305ab5/F0.jpg", "media/40f93ef1789faf305ab5/F1.jpg", "media/40f93ef1789faf305ab5/F2.jpg", "media/40f93ef1789faf305ab5/F3.jpg", "media/40f93ef1789faf305ab5/F4.jpg", "media/40f93ef1789faf305ab5/F5.jpg", "media/40f93ef1789faf305ab5/F6.jpg", "media/40f93ef1789faf305ab5/F7.jpg", "media/40f93ef1789faf305ab5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person in dark clothing sledding down a snowy road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "08d214dc4d975e994d5a792994db0ba0b8257429b0a2531bfaf9cd5e1b0ad8d4"}
{"qa_id": "uvo__-Z1X9B-F04Q__object_world_motion_direction__00", "video_id": "uvo__-Z1X9B-F04Q", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person wearing a yellow high-visibility vest and orange pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "down", "up", "forward"], "frame_paths": ["media/324019d9cbad95b31cac/F0.jpg", "media/324019d9cbad95b31cac/F1.jpg", "media/324019d9cbad95b31cac/F2.jpg", "media/324019d9cbad95b31cac/F3.jpg", "media/324019d9cbad95b31cac/F4.jpg", "media/324019d9cbad95b31cac/F5.jpg", "media/324019d9cbad95b31cac/F6.jpg", "media/324019d9cbad95b31cac/F7.jpg", "media/324019d9cbad95b31cac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person wearing a yellow high-visibility vest and orange pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d480cf1d0d489b129b6d092f4894db505ccb9db44b3083cd5b98a5c53e8d69d"}
{"qa_id": "uvo__-Z7SlpjF3cE__object_world_motion_direction__00", "video_id": "uvo__-Z7SlpjF3cE", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (child in purple shirt) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "down", "up", "forward", "backward", "right"], "frame_paths": ["media/e224f21c3b8fde442c85/F0.jpg", "media/e224f21c3b8fde442c85/F1.jpg", "media/e224f21c3b8fde442c85/F2.jpg", "media/e224f21c3b8fde442c85/F3.jpg", "media/e224f21c3b8fde442c85/F4.jpg", "media/e224f21c3b8fde442c85/F5.jpg", "media/e224f21c3b8fde442c85/F6.jpg", "media/e224f21c3b8fde442c85/F7.jpg", "media/e224f21c3b8fde442c85/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (child in purple shirt) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. up\nD. forward\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e92232ca23ce2c33c4d60e69df1cc886263126dc7c1f640a576a51b81f7396dd"}
{"qa_id": "uvo__-Z7SlpjF3cE__object_pair_distance_change__01", "video_id": "uvo__-Z7SlpjF3cE", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (child in purple shirt) and object B (man with short dark hair) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/294da3d55001fd767084/F0.jpg", "media/294da3d55001fd767084/F1.jpg", "media/294da3d55001fd767084/F2.jpg", "media/294da3d55001fd767084/F3.jpg", "media/294da3d55001fd767084/F4.jpg", "media/294da3d55001fd767084/F5.jpg", "media/294da3d55001fd767084/F6.jpg", "media/294da3d55001fd767084/F7.jpg", "media/294da3d55001fd767084/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (child in purple shirt) and object B (man with short dark hair) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f35715383908f7f212254eafd85d4f818a5906d9b16c415d507b75fcd07ee46"}
{"qa_id": "uvo__-Z7SlpjF3cE__object_travel_distance_comparison__02", "video_id": "uvo__-Z7SlpjF3cE", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in purple shirt) or B (man with short dark hair)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/294da3d55001fd767084/F0.jpg", "media/294da3d55001fd767084/F1.jpg", "media/294da3d55001fd767084/F2.jpg", "media/294da3d55001fd767084/F3.jpg", "media/294da3d55001fd767084/F4.jpg", "media/294da3d55001fd767084/F5.jpg", "media/294da3d55001fd767084/F6.jpg", "media/294da3d55001fd767084/F7.jpg", "media/294da3d55001fd767084/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (child in purple shirt) or B (man with short dark hair)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb1b28c428a07ea6dd445b1fbbe45df5246da385c034323879ef9ca7b485c849"}
{"qa_id": "uvo__-_srAE4uEpo__camera_object_distance_change__00", "video_id": "uvo__-_srAE4uEpo", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did object A (person in blue jacket and jeans) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/4b2e4dc23a0743d74aa0/F0.jpg", "media/4b2e4dc23a0743d74aa0/F1.jpg", "media/4b2e4dc23a0743d74aa0/F2.jpg", "media/4b2e4dc23a0743d74aa0/F3.jpg", "media/4b2e4dc23a0743d74aa0/F4.jpg", "media/4b2e4dc23a0743d74aa0/F5.jpg", "media/4b2e4dc23a0743d74aa0/F6.jpg", "media/4b2e4dc23a0743d74aa0/F7.jpg", "media/4b2e4dc23a0743d74aa0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person in blue jacket and jeans) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e4ced0ed5bd4b4ffefa0ff6c35ff49d825a7074f78cf12d25c4b5aeefed8622"}
{"qa_id": "uvo__-_srAE4uEpo__object_travel_distance_comparison__01", "video_id": "uvo__-_srAE4uEpo", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue jacket and jeans) or B (person in camouflage uniform)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/0e9037a8b7059f6b83b1/F0.jpg", "media/0e9037a8b7059f6b83b1/F1.jpg", "media/0e9037a8b7059f6b83b1/F2.jpg", "media/0e9037a8b7059f6b83b1/F3.jpg", "media/0e9037a8b7059f6b83b1/F4.jpg", "media/0e9037a8b7059f6b83b1/F5.jpg", "media/0e9037a8b7059f6b83b1/F6.jpg", "media/0e9037a8b7059f6b83b1/F7.jpg", "media/0e9037a8b7059f6b83b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue jacket and jeans) or B (person in camouflage uniform)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4131dcc888d7b048bba55f85c9b37b3d4d544bf2fb44b5569c4128bc5b095654"}
{"qa_id": "uvo__-acZ6mX_-nA__object_pair_distance_change__02", "video_id": "uvo__-acZ6mX_-nA", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did the 3D distance between object A (person in orange jacket) and object B (person in dark jacket) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/36988bbe96292ec22463/F0.jpg", "media/36988bbe96292ec22463/F1.jpg", "media/36988bbe96292ec22463/F2.jpg", "media/36988bbe96292ec22463/F3.jpg", "media/36988bbe96292ec22463/F4.jpg", "media/36988bbe96292ec22463/F5.jpg", "media/36988bbe96292ec22463/F6.jpg", "media/36988bbe96292ec22463/F7.jpg", "media/36988bbe96292ec22463/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (person in orange jacket) and object B (person in dark jacket) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84403a20ef973dbcc686dca4574f27448a8b2f213ea0e5a37a0886d12da94255"}
{"qa_id": "uvo__-b2fhk6ntm4__camera_translation_direction__00", "video_id": "uvo__-b2fhk6ntm4", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "down", "up", "forward", "left", "backward"], "frame_paths": ["media/2b66073a3970ec6b4c4d/F0.jpg", "media/2b66073a3970ec6b4c4d/F1.jpg", "media/2b66073a3970ec6b4c4d/F2.jpg", "media/2b66073a3970ec6b4c4d/F3.jpg", "media/2b66073a3970ec6b4c4d/F4.jpg", "media/2b66073a3970ec6b4c4d/F5.jpg", "media/2b66073a3970ec6b4c4d/F6.jpg", "media/2b66073a3970ec6b4c4d/F7.jpg", "media/2b66073a3970ec6b4c4d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. up\nD. forward\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9f6f2508b604654179535580c2c8ed68c555d7c2ad03c8379ab94124a33cccba"}
{"qa_id": "uvo__-b2fhk6ntm4__camera_view_direction_change__01", "video_id": "uvo__-b2fhk6ntm4", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/2b66073a3970ec6b4c4d/F0.jpg", "media/2b66073a3970ec6b4c4d/F1.jpg", "media/2b66073a3970ec6b4c4d/F2.jpg", "media/2b66073a3970ec6b4c4d/F3.jpg", "media/2b66073a3970ec6b4c4d/F4.jpg", "media/2b66073a3970ec6b4c4d/F5.jpg", "media/2b66073a3970ec6b4c4d/F6.jpg", "media/2b66073a3970ec6b4c4d/F7.jpg", "media/2b66073a3970ec6b4c4d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "50d4c5835a8c0b416e14c64b12cfc2d7486d1e2b0dc4e3f9c49d8962a91de366"}
{"qa_id": "uvo__-cUGz7HAGBg__object_world_motion_direction__00", "video_id": "uvo__-cUGz7HAGBg", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (camel with red saddle and rider) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "left", "up", "right"], "frame_paths": ["media/fa9d1ef56864f419279c/F0.jpg", "media/fa9d1ef56864f419279c/F1.jpg", "media/fa9d1ef56864f419279c/F2.jpg", "media/fa9d1ef56864f419279c/F3.jpg", "media/fa9d1ef56864f419279c/F4.jpg", "media/fa9d1ef56864f419279c/F5.jpg", "media/fa9d1ef56864f419279c/F6.jpg", "media/fa9d1ef56864f419279c/F7.jpg", "media/fa9d1ef56864f419279c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (camel with red saddle and rider) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9bd7a3a4478818ffd0bcd6d2448cfe9c25ff92349e09c93d245c42b099736efa"}
{"qa_id": "uvo__-cUGz7HAGBg__camera_object_distance_change__01", "video_id": "uvo__-cUGz7HAGBg", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did object A (camel with red saddle and rider) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fa9d1ef56864f419279c/F0.jpg", "media/fa9d1ef56864f419279c/F1.jpg", "media/fa9d1ef56864f419279c/F2.jpg", "media/fa9d1ef56864f419279c/F3.jpg", "media/fa9d1ef56864f419279c/F4.jpg", "media/fa9d1ef56864f419279c/F5.jpg", "media/fa9d1ef56864f419279c/F6.jpg", "media/fa9d1ef56864f419279c/F7.jpg", "media/fa9d1ef56864f419279c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (camel with red saddle and rider) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3ffe57d622eebbe63b6f2f2d790abbcad75e2f8af944399706caedad56e41bc"}
{"qa_id": "uvo__-cUGz7HAGBg__object_world_motion_direction__02", "video_id": "uvo__-cUGz7HAGBg", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (person in dark clothing standing near the camel) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "up", "down"], "frame_paths": ["media/a253e1f99b9a9508bc7d/F0.jpg", "media/a253e1f99b9a9508bc7d/F1.jpg", "media/a253e1f99b9a9508bc7d/F2.jpg", "media/a253e1f99b9a9508bc7d/F3.jpg", "media/a253e1f99b9a9508bc7d/F4.jpg", "media/a253e1f99b9a9508bc7d/F5.jpg", "media/a253e1f99b9a9508bc7d/F6.jpg", "media/a253e1f99b9a9508bc7d/F7.jpg", "media/a253e1f99b9a9508bc7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (person in dark clothing standing near the camel) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "94115be8b3d1ca7d64a34c7992af8d613ca55482cf313f57c11895a6598550f9"}
{"qa_id": "uvo__-eYXPoMyrXw__camera_view_direction_change__00", "video_id": "uvo__-eYXPoMyrXw", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/4a6dacf9672f5df08e37/F0.jpg", "media/4a6dacf9672f5df08e37/F1.jpg", "media/4a6dacf9672f5df08e37/F2.jpg", "media/4a6dacf9672f5df08e37/F3.jpg", "media/4a6dacf9672f5df08e37/F4.jpg", "media/4a6dacf9672f5df08e37/F5.jpg", "media/4a6dacf9672f5df08e37/F6.jpg", "media/4a6dacf9672f5df08e37/F7.jpg", "media/4a6dacf9672f5df08e37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af75473fbfd63923e7a7dd9d31a1213d36f19af8df543626bae3d6af8373e9d0"}
{"qa_id": "uvo__-fTmHyOG-CY__camera_translation_direction__00", "video_id": "uvo__-fTmHyOG-CY", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "down", "forward", "right"], "frame_paths": ["media/d540f98a16badeee2315/F0.jpg", "media/d540f98a16badeee2315/F1.jpg", "media/d540f98a16badeee2315/F2.jpg", "media/d540f98a16badeee2315/F3.jpg", "media/d540f98a16badeee2315/F4.jpg", "media/d540f98a16badeee2315/F5.jpg", "media/d540f98a16badeee2315/F6.jpg", "media/d540f98a16badeee2315/F7.jpg", "media/d540f98a16badeee2315/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2efd5607105f39a875a5f0876f896d746b93dd1aeb714e47da5972894a21c0c3"}
{"qa_id": "uvo__-fTmHyOG-CY__camera_view_direction_change__01", "video_id": "uvo__-fTmHyOG-CY", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/d540f98a16badeee2315/F0.jpg", "media/d540f98a16badeee2315/F1.jpg", "media/d540f98a16badeee2315/F2.jpg", "media/d540f98a16badeee2315/F3.jpg", "media/d540f98a16badeee2315/F4.jpg", "media/d540f98a16badeee2315/F5.jpg", "media/d540f98a16badeee2315/F6.jpg", "media/d540f98a16badeee2315/F7.jpg", "media/d540f98a16badeee2315/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a33bc8a883c2e465b120550c83d7462e1e26103c5ead8a6bedd8140b0a0fd32"}
{"qa_id": "uvo__-gcoV6rYgyE__camera_translation_direction__00", "video_id": "uvo__-gcoV6rYgyE", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "backward", "up", "down"], "frame_paths": ["media/5219b49e4a4fac33c713/F0.jpg", "media/5219b49e4a4fac33c713/F1.jpg", "media/5219b49e4a4fac33c713/F2.jpg", "media/5219b49e4a4fac33c713/F3.jpg", "media/5219b49e4a4fac33c713/F4.jpg", "media/5219b49e4a4fac33c713/F5.jpg", "media/5219b49e4a4fac33c713/F6.jpg", "media/5219b49e4a4fac33c713/F7.jpg", "media/5219b49e4a4fac33c713/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b45d7a64395dcc53d45d64fe1e5cf6e26976307859d5c0e1d8ebb7bb35a4c798"}
{"qa_id": "uvo__-gcoV6rYgyE__camera_view_direction_change__01", "video_id": "uvo__-gcoV6rYgyE", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/5219b49e4a4fac33c713/F0.jpg", "media/5219b49e4a4fac33c713/F1.jpg", "media/5219b49e4a4fac33c713/F2.jpg", "media/5219b49e4a4fac33c713/F3.jpg", "media/5219b49e4a4fac33c713/F4.jpg", "media/5219b49e4a4fac33c713/F5.jpg", "media/5219b49e4a4fac33c713/F6.jpg", "media/5219b49e4a4fac33c713/F7.jpg", "media/5219b49e4a4fac33c713/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3cefcc9233d0a5d863cee36e81831a84b94f9dcfe3d5edd00e87bdfbda06d9eb"}
{"qa_id": "uvo__-ge-jtX1Ik0__object_world_motion_direction__00", "video_id": "uvo__-ge-jtX1Ik0", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in orange jacket skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/b96f840ca0261fde2215/F0.jpg", "media/b96f840ca0261fde2215/F1.jpg", "media/b96f840ca0261fde2215/F2.jpg", "media/b96f840ca0261fde2215/F3.jpg", "media/b96f840ca0261fde2215/F4.jpg", "media/b96f840ca0261fde2215/F5.jpg", "media/b96f840ca0261fde2215/F6.jpg", "media/b96f840ca0261fde2215/F7.jpg", "media/b96f840ca0261fde2215/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in orange jacket skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "760a621496a68b4fd99e10f18e107bd5f42fc1336c53c16554536de73314b6a3"}
{"qa_id": "uvo__-ge-jtX1Ik0__object_world_motion_direction__01", "video_id": "uvo__-ge-jtX1Ik0", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in dark jacket skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "up", "left", "forward"], "frame_paths": ["media/89ff941f5d00d480183d/F0.jpg", "media/89ff941f5d00d480183d/F1.jpg", "media/89ff941f5d00d480183d/F2.jpg", "media/89ff941f5d00d480183d/F3.jpg", "media/89ff941f5d00d480183d/F4.jpg", "media/89ff941f5d00d480183d/F5.jpg", "media/89ff941f5d00d480183d/F6.jpg", "media/89ff941f5d00d480183d/F7.jpg", "media/89ff941f5d00d480183d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in dark jacket skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "14e748cdb0969714ece89eef30dd78cf142f5ba1c5b15b6d45aea42502b7024a"}
{"qa_id": "uvo__-heNc38TpNI__camera_view_direction_change__00", "video_id": "uvo__-heNc38TpNI", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/64681f1d45d2e7447ce5/F0.jpg", "media/64681f1d45d2e7447ce5/F1.jpg", "media/64681f1d45d2e7447ce5/F2.jpg", "media/64681f1d45d2e7447ce5/F3.jpg", "media/64681f1d45d2e7447ce5/F4.jpg", "media/64681f1d45d2e7447ce5/F5.jpg", "media/64681f1d45d2e7447ce5/F6.jpg", "media/64681f1d45d2e7447ce5/F7.jpg", "media/64681f1d45d2e7447ce5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7cb4feef62c3d9efd6762a749f2132199fae31e2272a2ff8d2b33844b5eac135"}
{"qa_id": "uvo__-hl1mBY0_Yo__object_world_motion_direction__00", "video_id": "uvo__-hl1mBY0_Yo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark clothing standing on the ice) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "up", "backward", "forward"], "frame_paths": ["media/5efa0edd26d42f0810cb/F0.jpg", "media/5efa0edd26d42f0810cb/F1.jpg", "media/5efa0edd26d42f0810cb/F2.jpg", "media/5efa0edd26d42f0810cb/F3.jpg", "media/5efa0edd26d42f0810cb/F4.jpg", "media/5efa0edd26d42f0810cb/F5.jpg", "media/5efa0edd26d42f0810cb/F6.jpg", "media/5efa0edd26d42f0810cb/F7.jpg", "media/5efa0edd26d42f0810cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark clothing standing on the ice) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa1098e44d1ad802cc1765eef49f07c095081f448b1c538a10e8d29a532c1e93"}
{"qa_id": "uvo__-hl1mBY0_Yo__camera_object_distance_change__01", "video_id": "uvo__-hl1mBY0_Yo", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did object A (person in dark clothing standing on the ice) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5efa0edd26d42f0810cb/F0.jpg", "media/5efa0edd26d42f0810cb/F1.jpg", "media/5efa0edd26d42f0810cb/F2.jpg", "media/5efa0edd26d42f0810cb/F3.jpg", "media/5efa0edd26d42f0810cb/F4.jpg", "media/5efa0edd26d42f0810cb/F5.jpg", "media/5efa0edd26d42f0810cb/F6.jpg", "media/5efa0edd26d42f0810cb/F7.jpg", "media/5efa0edd26d42f0810cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (person in dark clothing standing on the ice) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb0b7665fdeb6824c53cb333f3848b419e0ff4d6153c7951a5c3d3b1b8428828"}
{"qa_id": "uvo__-hl1mBY0_Yo__object_pair_distance_change__02", "video_id": "uvo__-hl1mBY0_Yo", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did the 3D distance between object A (person in dark clothing standing on the ice) and object B (person in a brown jacket and dark pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/70de6854666496a7e875/F0.jpg", "media/70de6854666496a7e875/F1.jpg", "media/70de6854666496a7e875/F2.jpg", "media/70de6854666496a7e875/F3.jpg", "media/70de6854666496a7e875/F4.jpg", "media/70de6854666496a7e875/F5.jpg", "media/70de6854666496a7e875/F6.jpg", "media/70de6854666496a7e875/F7.jpg", "media/70de6854666496a7e875/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (person in dark clothing standing on the ice) and object B (person in a brown jacket and dark pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0ef6439e50f111302483abe3c484e559e49a89fe2dbe07a6b60948292baa34cc"}
{"qa_id": "uvo__-hzfAaWtzYw__object_world_motion_direction__00", "video_id": "uvo__-hzfAaWtzYw", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (person in camouflage uniform walking in a line) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "right", "up", "left"], "frame_paths": ["media/74fa87295b64e242b5df/F0.jpg", "media/74fa87295b64e242b5df/F1.jpg", "media/74fa87295b64e242b5df/F2.jpg", "media/74fa87295b64e242b5df/F3.jpg", "media/74fa87295b64e242b5df/F4.jpg", "media/74fa87295b64e242b5df/F5.jpg", "media/74fa87295b64e242b5df/F6.jpg", "media/74fa87295b64e242b5df/F7.jpg", "media/74fa87295b64e242b5df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (person in camouflage uniform walking in a line) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "876171ca27f1672087451e63fa04df1f4156794171200eb7c1087f3d1416af3e"}
{"qa_id": "uvo__-hzfAaWtzYw__camera_object_distance_change__01", "video_id": "uvo__-hzfAaWtzYw", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (person in camouflage uniform walking in a line) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/74fa87295b64e242b5df/F0.jpg", "media/74fa87295b64e242b5df/F1.jpg", "media/74fa87295b64e242b5df/F2.jpg", "media/74fa87295b64e242b5df/F3.jpg", "media/74fa87295b64e242b5df/F4.jpg", "media/74fa87295b64e242b5df/F5.jpg", "media/74fa87295b64e242b5df/F6.jpg", "media/74fa87295b64e242b5df/F7.jpg", "media/74fa87295b64e242b5df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (person in camouflage uniform walking in a line) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a24f25dbe85d576f59952333770e3ffdada0577197cc3210185f5bd2281bc304"}
{"qa_id": "uvo__-jPBBLagPv4__camera_view_direction_change__00", "video_id": "uvo__-jPBBLagPv4", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/e15a4de4ebef2c91fc39/F0.jpg", "media/e15a4de4ebef2c91fc39/F1.jpg", "media/e15a4de4ebef2c91fc39/F2.jpg", "media/e15a4de4ebef2c91fc39/F3.jpg", "media/e15a4de4ebef2c91fc39/F4.jpg", "media/e15a4de4ebef2c91fc39/F5.jpg", "media/e15a4de4ebef2c91fc39/F6.jpg", "media/e15a4de4ebef2c91fc39/F7.jpg", "media/e15a4de4ebef2c91fc39/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53e2b83a88fe2eec3cdec99b6f57c27ac0474645f750fa59a38cf69cf5470b3"}
{"qa_id": "uvo__-jaztOANmrw__camera_view_direction_change__00", "video_id": "uvo__-jaztOANmrw", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/ea981a613d59b59332ae/F0.jpg", "media/ea981a613d59b59332ae/F1.jpg", "media/ea981a613d59b59332ae/F2.jpg", "media/ea981a613d59b59332ae/F3.jpg", "media/ea981a613d59b59332ae/F4.jpg", "media/ea981a613d59b59332ae/F5.jpg", "media/ea981a613d59b59332ae/F6.jpg", "media/ea981a613d59b59332ae/F7.jpg", "media/ea981a613d59b59332ae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8ce28c625bc67e3a214958c9c5377661f26f63aef3827bb443d619a99da34135"}
{"qa_id": "uvo__-jskAtCipd8__object_travel_distance_comparison__00", "video_id": "uvo__-jskAtCipd8", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (referee in black and white striped shirt and white cap) or B (football player in black jersey with number 8 and pink sleeve)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/720f8107c4878a126e97/F0.jpg", "media/720f8107c4878a126e97/F1.jpg", "media/720f8107c4878a126e97/F2.jpg", "media/720f8107c4878a126e97/F3.jpg", "media/720f8107c4878a126e97/F4.jpg", "media/720f8107c4878a126e97/F5.jpg", "media/720f8107c4878a126e97/F6.jpg", "media/720f8107c4878a126e97/F7.jpg", "media/720f8107c4878a126e97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (referee in black and white striped shirt and white cap) or B (football player in black jersey with number 8 and pink sleeve)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "926fd9635ce9a8060f12da91113d6118253d1b3a6312230e361161aa3c48125a"}
{"qa_id": "uvo__-k1jF9CkQ2A__object_world_motion_direction__00", "video_id": "uvo__-k1jF9CkQ2A", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person with short dark hair wearing a light grey t-shirt, viewed from behind) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/c50828200bcb10941c6c/F0.jpg", "media/c50828200bcb10941c6c/F1.jpg", "media/c50828200bcb10941c6c/F2.jpg", "media/c50828200bcb10941c6c/F3.jpg", "media/c50828200bcb10941c6c/F4.jpg", "media/c50828200bcb10941c6c/F5.jpg", "media/c50828200bcb10941c6c/F6.jpg", "media/c50828200bcb10941c6c/F7.jpg", "media/c50828200bcb10941c6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person with short dark hair wearing a light grey t-shirt, viewed from behind) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d64e02e52215bd58ea0d243397ad8248ea90f3fa233a2358ed3a81c5b767b91"}
{"qa_id": "uvo__-kZa3gn6ojw__camera_view_direction_change__00", "video_id": "uvo__-kZa3gn6ojw", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/987bace4ee69483cabf0/F0.jpg", "media/987bace4ee69483cabf0/F1.jpg", "media/987bace4ee69483cabf0/F2.jpg", "media/987bace4ee69483cabf0/F3.jpg", "media/987bace4ee69483cabf0/F4.jpg", "media/987bace4ee69483cabf0/F5.jpg", "media/987bace4ee69483cabf0/F6.jpg", "media/987bace4ee69483cabf0/F7.jpg", "media/987bace4ee69483cabf0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "acd59f5999362d81d16fc7a769204a3775b0696b58f897bc7e67bc16ca07099b"}
{"qa_id": "uvo__-kZa3gn6ojw__camera_object_distance_change__01", "video_id": "uvo__-kZa3gn6ojw", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, did object A (skier in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/bca1fbf9995f2800839b/F0.jpg", "media/bca1fbf9995f2800839b/F1.jpg", "media/bca1fbf9995f2800839b/F2.jpg", "media/bca1fbf9995f2800839b/F3.jpg", "media/bca1fbf9995f2800839b/F4.jpg", "media/bca1fbf9995f2800839b/F5.jpg", "media/bca1fbf9995f2800839b/F6.jpg", "media/bca1fbf9995f2800839b/F7.jpg", "media/bca1fbf9995f2800839b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (skier in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "127c91835f3d5ba8635bd225238eec19b9e7db387d6c22a82c09bd36f4712846"}
{"qa_id": "uvo__-kd_rdZ9Xaw__object_world_motion_direction__00", "video_id": "uvo__-kd_rdZ9Xaw", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in black shirt riding a horse) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "right", "down"], "frame_paths": ["media/4a2f16d38d2c02fcda86/F0.jpg", "media/4a2f16d38d2c02fcda86/F1.jpg", "media/4a2f16d38d2c02fcda86/F2.jpg", "media/4a2f16d38d2c02fcda86/F3.jpg", "media/4a2f16d38d2c02fcda86/F4.jpg", "media/4a2f16d38d2c02fcda86/F5.jpg", "media/4a2f16d38d2c02fcda86/F6.jpg", "media/4a2f16d38d2c02fcda86/F7.jpg", "media/4a2f16d38d2c02fcda86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in black shirt riding a horse) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13e51a7b45e1d02bb1f9a37d7b93fb946b2953a9e02b8a22e8e30be68a1151c2"}
{"qa_id": "uvo__-kd_rdZ9Xaw__camera_object_distance_change__01", "video_id": "uvo__-kd_rdZ9Xaw", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object A (person in black shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/4a2f16d38d2c02fcda86/F0.jpg", "media/4a2f16d38d2c02fcda86/F1.jpg", "media/4a2f16d38d2c02fcda86/F2.jpg", "media/4a2f16d38d2c02fcda86/F3.jpg", "media/4a2f16d38d2c02fcda86/F4.jpg", "media/4a2f16d38d2c02fcda86/F5.jpg", "media/4a2f16d38d2c02fcda86/F6.jpg", "media/4a2f16d38d2c02fcda86/F7.jpg", "media/4a2f16d38d2c02fcda86/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person in black shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "54e2c9a8a24790404dd7607a7d0c3d61eeb6d1ca322eaf46c948ede43abf54aa"}
{"qa_id": "uvo__-mDRf2N08EI__camera_translation_direction__00", "video_id": "uvo__-mDRf2N08EI", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "forward", "down", "up"], "frame_paths": ["media/9bcef05e35512f1535d0/F0.jpg", "media/9bcef05e35512f1535d0/F1.jpg", "media/9bcef05e35512f1535d0/F2.jpg", "media/9bcef05e35512f1535d0/F3.jpg", "media/9bcef05e35512f1535d0/F4.jpg", "media/9bcef05e35512f1535d0/F5.jpg", "media/9bcef05e35512f1535d0/F6.jpg", "media/9bcef05e35512f1535d0/F7.jpg", "media/9bcef05e35512f1535d0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "32968ab4e177e5a6de717b4d31e37c680451bd1913804e76faf013e840480ffb"}
{"qa_id": "uvo__-mDRf2N08EI__object_world_motion_direction__01", "video_id": "uvo__-mDRf2N08EI", "task": "object_world_motion_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (brown horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "left", "up", "backward"], "frame_paths": ["media/e96d61198d705d019dd3/F0.jpg", "media/e96d61198d705d019dd3/F1.jpg", "media/e96d61198d705d019dd3/F2.jpg", "media/e96d61198d705d019dd3/F3.jpg", "media/e96d61198d705d019dd3/F4.jpg", "media/e96d61198d705d019dd3/F5.jpg", "media/e96d61198d705d019dd3/F6.jpg", "media/e96d61198d705d019dd3/F7.jpg", "media/e96d61198d705d019dd3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (brown horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87d67c76f2cea929f921c5bb3dc20d2fd77a39992a8bba14192a304a7cfbb41f"}
{"qa_id": "uvo__-mDRf2N08EI__camera_object_distance_change__02", "video_id": "uvo__-mDRf2N08EI", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, did object A (brown horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/e96d61198d705d019dd3/F0.jpg", "media/e96d61198d705d019dd3/F1.jpg", "media/e96d61198d705d019dd3/F2.jpg", "media/e96d61198d705d019dd3/F3.jpg", "media/e96d61198d705d019dd3/F4.jpg", "media/e96d61198d705d019dd3/F5.jpg", "media/e96d61198d705d019dd3/F6.jpg", "media/e96d61198d705d019dd3/F7.jpg", "media/e96d61198d705d019dd3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (brown horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a27920dedfd63146983012bb76093b3b29520569d325373c2a6a6517dbd16374"}
{"qa_id": "uvo__-mDRf2N08EI__object_pair_distance_change__03", "video_id": "uvo__-mDRf2N08EI", "task": "object_pair_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, did the 3D distance between object A (brown horse) and object B (dark horse) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/99eed23c17a960e590a9/F0.jpg", "media/99eed23c17a960e590a9/F1.jpg", "media/99eed23c17a960e590a9/F2.jpg", "media/99eed23c17a960e590a9/F3.jpg", "media/99eed23c17a960e590a9/F4.jpg", "media/99eed23c17a960e590a9/F5.jpg", "media/99eed23c17a960e590a9/F6.jpg", "media/99eed23c17a960e590a9/F7.jpg", "media/99eed23c17a960e590a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did the 3D distance between object A (brown horse) and object B (dark horse) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "152bcba328961dba9e662af75e382dac66812f06fbed78a2c7398aeecfa98ad3"}
{"qa_id": "uvo__-mDRf2N08EI__object_travel_distance_comparison__04", "video_id": "uvo__-mDRf2N08EI", "task": "object_travel_distance_comparison", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse) or B (dark horse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/99eed23c17a960e590a9/F0.jpg", "media/99eed23c17a960e590a9/F1.jpg", "media/99eed23c17a960e590a9/F2.jpg", "media/99eed23c17a960e590a9/F3.jpg", "media/99eed23c17a960e590a9/F4.jpg", "media/99eed23c17a960e590a9/F5.jpg", "media/99eed23c17a960e590a9/F6.jpg", "media/99eed23c17a960e590a9/F7.jpg", "media/99eed23c17a960e590a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse) or B (dark horse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "166f6d9ed3020ab6c752e46c60fc31cfcd45a8934c782c0175a3adcb8292859d"}
{"qa_id": "uvo__-nHjW0ifdi8__camera_view_direction_change__00", "video_id": "uvo__-nHjW0ifdi8", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/21b80123cb3c5e0ab4c0/F0.jpg", "media/21b80123cb3c5e0ab4c0/F1.jpg", "media/21b80123cb3c5e0ab4c0/F2.jpg", "media/21b80123cb3c5e0ab4c0/F3.jpg", "media/21b80123cb3c5e0ab4c0/F4.jpg", "media/21b80123cb3c5e0ab4c0/F5.jpg", "media/21b80123cb3c5e0ab4c0/F6.jpg", "media/21b80123cb3c5e0ab4c0/F7.jpg", "media/21b80123cb3c5e0ab4c0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e94aa5aa390b996348d3ba18f896bc695065272669917f1a4e0987378a87ac1d"}
{"qa_id": "uvo__-nbHB0eCJGQ__object_world_motion_direction__00", "video_id": "uvo__-nbHB0eCJGQ", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (athlete in blue uniform running on track) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "forward", "backward", "down"], "frame_paths": ["media/f778c2ec56a0562e7d3f/F0.jpg", "media/f778c2ec56a0562e7d3f/F1.jpg", "media/f778c2ec56a0562e7d3f/F2.jpg", "media/f778c2ec56a0562e7d3f/F3.jpg", "media/f778c2ec56a0562e7d3f/F4.jpg", "media/f778c2ec56a0562e7d3f/F5.jpg", "media/f778c2ec56a0562e7d3f/F6.jpg", "media/f778c2ec56a0562e7d3f/F7.jpg", "media/f778c2ec56a0562e7d3f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (athlete in blue uniform running on track) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. forward\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9976b6aa64fb99d42d45ab61b0b7f8b2f2ab0829433b86c6f722270a42afbc9a"}
{"qa_id": "uvo__-nbHB0eCJGQ__camera_object_distance_change__01", "video_id": "uvo__-nbHB0eCJGQ", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object A (athlete in blue uniform running on track) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f778c2ec56a0562e7d3f/F0.jpg", "media/f778c2ec56a0562e7d3f/F1.jpg", "media/f778c2ec56a0562e7d3f/F2.jpg", "media/f778c2ec56a0562e7d3f/F3.jpg", "media/f778c2ec56a0562e7d3f/F4.jpg", "media/f778c2ec56a0562e7d3f/F5.jpg", "media/f778c2ec56a0562e7d3f/F6.jpg", "media/f778c2ec56a0562e7d3f/F7.jpg", "media/f778c2ec56a0562e7d3f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (athlete in blue uniform running on track) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "beb913187b2d6507eae5acedb5431afaa67fd019cab90ad62d58f778b4ffc3ff"}
{"qa_id": "uvo__-nt0YjZdboY__camera_translation_direction__00", "video_id": "uvo__-nt0YjZdboY", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "backward", "down", "right"], "frame_paths": ["media/cf8eb16830118f7bf7f2/F0.jpg", "media/cf8eb16830118f7bf7f2/F1.jpg", "media/cf8eb16830118f7bf7f2/F2.jpg", "media/cf8eb16830118f7bf7f2/F3.jpg", "media/cf8eb16830118f7bf7f2/F4.jpg", "media/cf8eb16830118f7bf7f2/F5.jpg", "media/cf8eb16830118f7bf7f2/F6.jpg", "media/cf8eb16830118f7bf7f2/F7.jpg", "media/cf8eb16830118f7bf7f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. backward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e8233833275428b5cc84ed97704073f831ab37253ebd062951afdd6cc12a22a"}
{"qa_id": "uvo__-nt0YjZdboY__object_world_motion_direction__01", "video_id": "uvo__-nt0YjZdboY", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (toddler in black shirt and light green shorts) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "forward", "up", "right", "down"], "frame_paths": ["media/146933c0d499eb6535e8/F0.jpg", "media/146933c0d499eb6535e8/F1.jpg", "media/146933c0d499eb6535e8/F2.jpg", "media/146933c0d499eb6535e8/F3.jpg", "media/146933c0d499eb6535e8/F4.jpg", "media/146933c0d499eb6535e8/F5.jpg", "media/146933c0d499eb6535e8/F6.jpg", "media/146933c0d499eb6535e8/F7.jpg", "media/146933c0d499eb6535e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (toddler in black shirt and light green shorts) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. forward\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2b5e17640f410081319d4818f1ee8599e9e888fc60057384346846eadefc9766"}
{"qa_id": "uvo__-nxElRBL0VU__camera_view_direction_change__00", "video_id": "uvo__-nxElRBL0VU", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/b3b8ab5f99128f051660/F0.jpg", "media/b3b8ab5f99128f051660/F1.jpg", "media/b3b8ab5f99128f051660/F2.jpg", "media/b3b8ab5f99128f051660/F3.jpg", "media/b3b8ab5f99128f051660/F4.jpg", "media/b3b8ab5f99128f051660/F5.jpg", "media/b3b8ab5f99128f051660/F6.jpg", "media/b3b8ab5f99128f051660/F7.jpg", "media/b3b8ab5f99128f051660/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bec691eba368f26b6ba745fdd6a3eb3939004e6d5b9e30ac931bbc88d0b17629"}
{"qa_id": "uvo__-pOpjuLHCqk__object_world_motion_direction__00", "video_id": "uvo__-pOpjuLHCqk", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (snowboarder wearing a red jacket and black helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "left", "down", "right"], "frame_paths": ["media/e5e511e6600a74513eaf/F0.jpg", "media/e5e511e6600a74513eaf/F1.jpg", "media/e5e511e6600a74513eaf/F2.jpg", "media/e5e511e6600a74513eaf/F3.jpg", "media/e5e511e6600a74513eaf/F4.jpg", "media/e5e511e6600a74513eaf/F5.jpg", "media/e5e511e6600a74513eaf/F6.jpg", "media/e5e511e6600a74513eaf/F7.jpg", "media/e5e511e6600a74513eaf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (snowboarder wearing a red jacket and black helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. left\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b8e8f11fceb1ca2475e062f144bdb1ce864b49934894624886b89d256f64856"}
{"qa_id": "uvo__-pOpjuLHCqk__camera_object_distance_change__01", "video_id": "uvo__-pOpjuLHCqk", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object A (snowboarder wearing a red jacket and black helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/e5e511e6600a74513eaf/F0.jpg", "media/e5e511e6600a74513eaf/F1.jpg", "media/e5e511e6600a74513eaf/F2.jpg", "media/e5e511e6600a74513eaf/F3.jpg", "media/e5e511e6600a74513eaf/F4.jpg", "media/e5e511e6600a74513eaf/F5.jpg", "media/e5e511e6600a74513eaf/F6.jpg", "media/e5e511e6600a74513eaf/F7.jpg", "media/e5e511e6600a74513eaf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (snowboarder wearing a red jacket and black helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8bd3cc040c655d036d077b3fce124c2b294fd00dec981cef85de560c534b5a8f"}
{"qa_id": "uvo__-pOpjuLHCqk__object_world_motion_direction__02", "video_id": "uvo__-pOpjuLHCqk", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (snowboarder wearing black patterned pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "right", "backward", "down", "left", "forward"], "frame_paths": ["media/ad81776acb3dd45a9884/F0.jpg", "media/ad81776acb3dd45a9884/F1.jpg", "media/ad81776acb3dd45a9884/F2.jpg", "media/ad81776acb3dd45a9884/F3.jpg", "media/ad81776acb3dd45a9884/F4.jpg", "media/ad81776acb3dd45a9884/F5.jpg", "media/ad81776acb3dd45a9884/F6.jpg", "media/ad81776acb3dd45a9884/F7.jpg", "media/ad81776acb3dd45a9884/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (snowboarder wearing black patterned pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. backward\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64ad58ad7fe810a80656319cb6982f2075e72fb17327bd3e5a8e462fc138ce88"}
{"qa_id": "uvo__-pOpjuLHCqk__camera_object_distance_change__03", "video_id": "uvo__-pOpjuLHCqk", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object B (snowboarder wearing black patterned pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ad81776acb3dd45a9884/F0.jpg", "media/ad81776acb3dd45a9884/F1.jpg", "media/ad81776acb3dd45a9884/F2.jpg", "media/ad81776acb3dd45a9884/F3.jpg", "media/ad81776acb3dd45a9884/F4.jpg", "media/ad81776acb3dd45a9884/F5.jpg", "media/ad81776acb3dd45a9884/F6.jpg", "media/ad81776acb3dd45a9884/F7.jpg", "media/ad81776acb3dd45a9884/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (snowboarder wearing black patterned pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6250d3b5716b8f76c766106d69387969c290e3138df4456fd27bbe98bccef3f5"}
{"qa_id": "uvo__-pUiDc8X-To__camera_translation_direction__00", "video_id": "uvo__-pUiDc8X-To", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "right", "up", "left", "down"], "frame_paths": ["media/46709e5fa8f9056f8148/F0.jpg", "media/46709e5fa8f9056f8148/F1.jpg", "media/46709e5fa8f9056f8148/F2.jpg", "media/46709e5fa8f9056f8148/F3.jpg", "media/46709e5fa8f9056f8148/F4.jpg", "media/46709e5fa8f9056f8148/F5.jpg", "media/46709e5fa8f9056f8148/F6.jpg", "media/46709e5fa8f9056f8148/F7.jpg", "media/46709e5fa8f9056f8148/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. right\nD. up\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "284aba559b7009298b38a1d9917d29c26885b8bbedc1f008901eb138a45fa803"}
{"qa_id": "uvo__-pUiDc8X-To__object_travel_distance_comparison__01", "video_id": "uvo__-pUiDc8X-To", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light-colored dog with a red harness) or B (child in a blue shirt and denim shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2d593cca9d74a6231212/F0.jpg", "media/2d593cca9d74a6231212/F1.jpg", "media/2d593cca9d74a6231212/F2.jpg", "media/2d593cca9d74a6231212/F3.jpg", "media/2d593cca9d74a6231212/F4.jpg", "media/2d593cca9d74a6231212/F5.jpg", "media/2d593cca9d74a6231212/F6.jpg", "media/2d593cca9d74a6231212/F7.jpg", "media/2d593cca9d74a6231212/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light-colored dog with a red harness) or B (child in a blue shirt and denim shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "992c50e37efd71e92cbad1dfab5a6276378fed02878453fc991f925eb6cad63e"}
{"qa_id": "uvo__-pWAEIT0HRo__camera_translation_direction__00", "video_id": "uvo__-pWAEIT0HRo", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "down", "forward"], "frame_paths": ["media/ac08562c7f51e2f659b5/F0.jpg", "media/ac08562c7f51e2f659b5/F1.jpg", "media/ac08562c7f51e2f659b5/F2.jpg", "media/ac08562c7f51e2f659b5/F3.jpg", "media/ac08562c7f51e2f659b5/F4.jpg", "media/ac08562c7f51e2f659b5/F5.jpg", "media/ac08562c7f51e2f659b5/F6.jpg", "media/ac08562c7f51e2f659b5/F7.jpg", "media/ac08562c7f51e2f659b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce5e32a00d9b5cdba723d9e9ef8d91dee5fd3680c27622c76e6bfe18437bee09"}
{"qa_id": "uvo__-u59xQTOcmY__camera_translation_direction__00", "video_id": "uvo__-u59xQTOcmY", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "up", "backward"], "frame_paths": ["media/5218e55e0608c561f59d/F0.jpg", "media/5218e55e0608c561f59d/F1.jpg", "media/5218e55e0608c561f59d/F2.jpg", "media/5218e55e0608c561f59d/F3.jpg", "media/5218e55e0608c561f59d/F4.jpg", "media/5218e55e0608c561f59d/F5.jpg", "media/5218e55e0608c561f59d/F6.jpg", "media/5218e55e0608c561f59d/F7.jpg", "media/5218e55e0608c561f59d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "412154081e34a496ae17af57280f8b73e2abd7b3795e190442dbb5b4c1c5f7e3"}
{"qa_id": "uvo__-ujBLQ7LF40__camera_view_direction_change__00", "video_id": "uvo__-ujBLQ7LF40", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/213a560ceeb95f1c2a89/F0.jpg", "media/213a560ceeb95f1c2a89/F1.jpg", "media/213a560ceeb95f1c2a89/F2.jpg", "media/213a560ceeb95f1c2a89/F3.jpg", "media/213a560ceeb95f1c2a89/F4.jpg", "media/213a560ceeb95f1c2a89/F5.jpg", "media/213a560ceeb95f1c2a89/F6.jpg", "media/213a560ceeb95f1c2a89/F7.jpg", "media/213a560ceeb95f1c2a89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e1b44109117f82af47bdb0b4bf158cae5c21da02933ade422dfbded4a0ab2db5"}
{"qa_id": "uvo__-vAOikxwtSI__camera_view_direction_change__00", "video_id": "uvo__-vAOikxwtSI", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/de838b000d931f67e521/F0.jpg", "media/de838b000d931f67e521/F1.jpg", "media/de838b000d931f67e521/F2.jpg", "media/de838b000d931f67e521/F3.jpg", "media/de838b000d931f67e521/F4.jpg", "media/de838b000d931f67e521/F5.jpg", "media/de838b000d931f67e521/F6.jpg", "media/de838b000d931f67e521/F7.jpg", "media/de838b000d931f67e521/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb2d3fb6ca745b37b06f7bb335d8e826a27d1c796182f53c3ba59327806b35c"}
{"qa_id": "uvo__-wHLbRfZPWM__camera_object_distance_change__01", "video_id": "uvo__-wHLbRfZPWM", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object A (person in black shirt leading the camel) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/56f929a1a44238e8ed9f/F0.jpg", "media/56f929a1a44238e8ed9f/F1.jpg", "media/56f929a1a44238e8ed9f/F2.jpg", "media/56f929a1a44238e8ed9f/F3.jpg", "media/56f929a1a44238e8ed9f/F4.jpg", "media/56f929a1a44238e8ed9f/F5.jpg", "media/56f929a1a44238e8ed9f/F6.jpg", "media/56f929a1a44238e8ed9f/F7.jpg", "media/56f929a1a44238e8ed9f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (person in black shirt leading the camel) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8561e889b3f09c72e84bc8fd62cedaa7e9f5dc11987dbb6378a2e5a123092113"}
{"qa_id": "uvo__-xuVfmYKSsE__camera_translation_direction__00", "video_id": "uvo__-xuVfmYKSsE", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "right", "forward", "up"], "frame_paths": ["media/bc22db835b1310f7a2b5/F0.jpg", "media/bc22db835b1310f7a2b5/F1.jpg", "media/bc22db835b1310f7a2b5/F2.jpg", "media/bc22db835b1310f7a2b5/F3.jpg", "media/bc22db835b1310f7a2b5/F4.jpg", "media/bc22db835b1310f7a2b5/F5.jpg", "media/bc22db835b1310f7a2b5/F6.jpg", "media/bc22db835b1310f7a2b5/F7.jpg", "media/bc22db835b1310f7a2b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55fc3c86083399d302226b0524be241db01b8be0c8cedc4e0b9b3d12ca8c325a"}
{"qa_id": "uvo__-xuVfmYKSsE__camera_view_direction_change__01", "video_id": "uvo__-xuVfmYKSsE", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/bc22db835b1310f7a2b5/F0.jpg", "media/bc22db835b1310f7a2b5/F1.jpg", "media/bc22db835b1310f7a2b5/F2.jpg", "media/bc22db835b1310f7a2b5/F3.jpg", "media/bc22db835b1310f7a2b5/F4.jpg", "media/bc22db835b1310f7a2b5/F5.jpg", "media/bc22db835b1310f7a2b5/F6.jpg", "media/bc22db835b1310f7a2b5/F7.jpg", "media/bc22db835b1310f7a2b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4fed381014b48b577da2106981c7e3c6598f72d2704258e39caf47112ebe01e3"}
{"qa_id": "uvo__-xuVfmYKSsE__camera_object_distance_change__02", "video_id": "uvo__-xuVfmYKSsE", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did object A (dark horse jumping over a fence) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fd115dde19cf8788c98c/F0.jpg", "media/fd115dde19cf8788c98c/F1.jpg", "media/fd115dde19cf8788c98c/F2.jpg", "media/fd115dde19cf8788c98c/F3.jpg", "media/fd115dde19cf8788c98c/F4.jpg", "media/fd115dde19cf8788c98c/F5.jpg", "media/fd115dde19cf8788c98c/F6.jpg", "media/fd115dde19cf8788c98c/F7.jpg", "media/fd115dde19cf8788c98c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (dark horse jumping over a fence) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e4ecff06bb3fc45178ca242633a47b35d4a689da2d5915f0007a6cbaab80ccf"}
{"qa_id": "uvo__-yhePQ6NFEw__object_world_motion_direction__00", "video_id": "uvo__-yhePQ6NFEw", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in dark shirt and orange pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "down", "left", "up"], "frame_paths": ["media/7c2c52331f8b79afc347/F0.jpg", "media/7c2c52331f8b79afc347/F1.jpg", "media/7c2c52331f8b79afc347/F2.jpg", "media/7c2c52331f8b79afc347/F3.jpg", "media/7c2c52331f8b79afc347/F4.jpg", "media/7c2c52331f8b79afc347/F5.jpg", "media/7c2c52331f8b79afc347/F6.jpg", "media/7c2c52331f8b79afc347/F7.jpg", "media/7c2c52331f8b79afc347/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in dark shirt and orange pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "012f7286ac1ada429804bc930ae5ac3dc5b0a6cc9639d46c3c33616c827b405a"}
{"qa_id": "uvo__-ytKvwrcXT4__camera_view_direction_change__00", "video_id": "uvo__-ytKvwrcXT4", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/f4c11d52dd4a75f62b94/F0.jpg", "media/f4c11d52dd4a75f62b94/F1.jpg", "media/f4c11d52dd4a75f62b94/F2.jpg", "media/f4c11d52dd4a75f62b94/F3.jpg", "media/f4c11d52dd4a75f62b94/F4.jpg", "media/f4c11d52dd4a75f62b94/F5.jpg", "media/f4c11d52dd4a75f62b94/F6.jpg", "media/f4c11d52dd4a75f62b94/F7.jpg", "media/f4c11d52dd4a75f62b94/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "uvo__-zjCHnN9SdQ__camera_view_direction_change__00", "video_id": "uvo__-zjCHnN9SdQ", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/c7bea2c1411d2de93e94/F0.jpg", "media/c7bea2c1411d2de93e94/F1.jpg", "media/c7bea2c1411d2de93e94/F2.jpg", "media/c7bea2c1411d2de93e94/F3.jpg", "media/c7bea2c1411d2de93e94/F4.jpg", "media/c7bea2c1411d2de93e94/F5.jpg", "media/c7bea2c1411d2de93e94/F6.jpg", "media/c7bea2c1411d2de93e94/F7.jpg", "media/c7bea2c1411d2de93e94/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc12d79511e7a4c2703fe469657077dfa5a1342963c60f9b378f06626186956d"}
{"qa_id": "uvo__-zmyDOwxEaE__object_pair_distance_change__00", "video_id": "uvo__-zmyDOwxEaE", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (person in white shirt and black cap riding a bicycle) and object B (person in black shirt and helmet riding a bicycle) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/5819bfb2fc354ac4ee18/F0.jpg", "media/5819bfb2fc354ac4ee18/F1.jpg", "media/5819bfb2fc354ac4ee18/F2.jpg", "media/5819bfb2fc354ac4ee18/F3.jpg", "media/5819bfb2fc354ac4ee18/F4.jpg", "media/5819bfb2fc354ac4ee18/F5.jpg", "media/5819bfb2fc354ac4ee18/F6.jpg", "media/5819bfb2fc354ac4ee18/F7.jpg", "media/5819bfb2fc354ac4ee18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (person in white shirt and black cap riding a bicycle) and object B (person in black shirt and helmet riding a bicycle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a25d2a4c7b2a4ef4ee54249046dd1afe2ded0c8af64c176ab5b93916a7c042c"}
{"qa_id": "uvo__-zmyDOwxEaE__object_travel_distance_comparison__01", "video_id": "uvo__-zmyDOwxEaE", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and black cap riding a bicycle) or B (person in black shirt and helmet riding a bicycle)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/5819bfb2fc354ac4ee18/F0.jpg", "media/5819bfb2fc354ac4ee18/F1.jpg", "media/5819bfb2fc354ac4ee18/F2.jpg", "media/5819bfb2fc354ac4ee18/F3.jpg", "media/5819bfb2fc354ac4ee18/F4.jpg", "media/5819bfb2fc354ac4ee18/F5.jpg", "media/5819bfb2fc354ac4ee18/F6.jpg", "media/5819bfb2fc354ac4ee18/F7.jpg", "media/5819bfb2fc354ac4ee18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and black cap riding a bicycle) or B (person in black shirt and helmet riding a bicycle)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e39ab1fcb0ad71d0446d8062306b1245b4d814835d0b9247796d151be9bda83"}
{"qa_id": "uvo__-zoJdjD91Uk__object_world_motion_direction__00", "video_id": "uvo__-zoJdjD91Uk", "task": "object_world_motion_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in dark clothing sitting in a boat) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "up", "down", "right", "forward"], "frame_paths": ["media/46a98b4059515e1c0269/F0.jpg", "media/46a98b4059515e1c0269/F1.jpg", "media/46a98b4059515e1c0269/F2.jpg", "media/46a98b4059515e1c0269/F3.jpg", "media/46a98b4059515e1c0269/F4.jpg", "media/46a98b4059515e1c0269/F5.jpg", "media/46a98b4059515e1c0269/F6.jpg", "media/46a98b4059515e1c0269/F7.jpg", "media/46a98b4059515e1c0269/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F8, in which dominant direction did object A (person in dark clothing sitting in a boat) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. up\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b86086aaf2406be1083719a866a6302eff2d3a93d05eb1cfb80aa6e7fd9f2ee1"}
{"qa_id": "uvo__0-PpgRaT3Es__object_world_motion_direction__00", "video_id": "uvo__0-PpgRaT3Es", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (woman with short hair walking, wearing a white top and black cardigan) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "down", "backward", "left"], "frame_paths": ["media/3510ddc1e057d29a8df9/F0.jpg", "media/3510ddc1e057d29a8df9/F1.jpg", "media/3510ddc1e057d29a8df9/F2.jpg", "media/3510ddc1e057d29a8df9/F3.jpg", "media/3510ddc1e057d29a8df9/F4.jpg", "media/3510ddc1e057d29a8df9/F5.jpg", "media/3510ddc1e057d29a8df9/F6.jpg", "media/3510ddc1e057d29a8df9/F7.jpg", "media/3510ddc1e057d29a8df9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (woman with short hair walking, wearing a white top and black cardigan) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. down\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d30d3e4294b3b2414938db13359381e38fc1faddb3c852447e134e438d5245a8"}
{"qa_id": "uvo__0-PpgRaT3Es__camera_object_distance_change__01", "video_id": "uvo__0-PpgRaT3Es", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did object B (woman with short hair walking, wearing a white top and black cardigan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/3510ddc1e057d29a8df9/F0.jpg", "media/3510ddc1e057d29a8df9/F1.jpg", "media/3510ddc1e057d29a8df9/F2.jpg", "media/3510ddc1e057d29a8df9/F3.jpg", "media/3510ddc1e057d29a8df9/F4.jpg", "media/3510ddc1e057d29a8df9/F5.jpg", "media/3510ddc1e057d29a8df9/F6.jpg", "media/3510ddc1e057d29a8df9/F7.jpg", "media/3510ddc1e057d29a8df9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (woman with short hair walking, wearing a white top and black cardigan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "18e4800a24ac366899a23e26f47c33b5aeed3dd669b455c6f3347d1de37058fb"}
{"qa_id": "uvo__0-PpgRaT3Es__object_pair_distance_change__02", "video_id": "uvo__0-PpgRaT3Es", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did the 3D distance between object A (man with long hair playing a cello, wearing a plaid shirt) and object B (woman with short hair walking, wearing a white top and black cardigan) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/4f0153269298e2355e0c/F0.jpg", "media/4f0153269298e2355e0c/F1.jpg", "media/4f0153269298e2355e0c/F2.jpg", "media/4f0153269298e2355e0c/F3.jpg", "media/4f0153269298e2355e0c/F4.jpg", "media/4f0153269298e2355e0c/F5.jpg", "media/4f0153269298e2355e0c/F6.jpg", "media/4f0153269298e2355e0c/F7.jpg", "media/4f0153269298e2355e0c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did the 3D distance between object A (man with long hair playing a cello, wearing a plaid shirt) and object B (woman with short hair walking, wearing a white top and black cardigan) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0f540ccb512c4e3133c8edb5ec1298a2abaae5f3574706c7fc945d842729f0f"}
{"qa_id": "uvo__0-PpgRaT3Es__object_travel_distance_comparison__03", "video_id": "uvo__0-PpgRaT3Es", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man with long hair playing a cello, wearing a plaid shirt) or B (woman with short hair walking, wearing a white top and black cardigan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/4f0153269298e2355e0c/F0.jpg", "media/4f0153269298e2355e0c/F1.jpg", "media/4f0153269298e2355e0c/F2.jpg", "media/4f0153269298e2355e0c/F3.jpg", "media/4f0153269298e2355e0c/F4.jpg", "media/4f0153269298e2355e0c/F5.jpg", "media/4f0153269298e2355e0c/F6.jpg", "media/4f0153269298e2355e0c/F7.jpg", "media/4f0153269298e2355e0c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g5"}, "requested_object_ids": ["g2", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man with long hair playing a cello, wearing a plaid shirt) or B (woman with short hair walking, wearing a white top and black cardigan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5da1044710a8d727910a066880357ae049e1b7ee9079baa0485cad4fb2973056"}
{"qa_id": "uvo__0-rDoto2S8o__camera_view_direction_change__00", "video_id": "uvo__0-rDoto2S8o", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/8c163ae299b2212908ad/F0.jpg", "media/8c163ae299b2212908ad/F1.jpg", "media/8c163ae299b2212908ad/F2.jpg", "media/8c163ae299b2212908ad/F3.jpg", "media/8c163ae299b2212908ad/F4.jpg", "media/8c163ae299b2212908ad/F5.jpg", "media/8c163ae299b2212908ad/F6.jpg", "media/8c163ae299b2212908ad/F7.jpg", "media/8c163ae299b2212908ad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ac0afaee5108d6c0556a29dab5077f25d0198ed35721598f3808e35adcc8aa4b"}
{"qa_id": "uvo__0-uAUzHjPBo__object_pair_distance_change__00", "video_id": "uvo__0-uAUzHjPBo", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F8, did the 3D distance between object A (person in orange shirt) and object B (person in black shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/2e546f67da17cbff8ffd/F0.jpg", "media/2e546f67da17cbff8ffd/F1.jpg", "media/2e546f67da17cbff8ffd/F2.jpg", "media/2e546f67da17cbff8ffd/F3.jpg", "media/2e546f67da17cbff8ffd/F4.jpg", "media/2e546f67da17cbff8ffd/F5.jpg", "media/2e546f67da17cbff8ffd/F6.jpg", "media/2e546f67da17cbff8ffd/F7.jpg", "media/2e546f67da17cbff8ffd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person in orange shirt) and object B (person in black shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e6041d437f8d4d8f156d9c66425d2218c4bbfa73bbb205b754c15ad301a9ac4"}
{"qa_id": "uvo__00HGQ6LqJZ8__object_world_motion_direction__00", "video_id": "uvo__00HGQ6LqJZ8", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in white shirt and purple shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "down", "up", "right", "left"], "frame_paths": ["media/b1fd067a65813f7f9ac5/F0.jpg", "media/b1fd067a65813f7f9ac5/F1.jpg", "media/b1fd067a65813f7f9ac5/F2.jpg", "media/b1fd067a65813f7f9ac5/F3.jpg", "media/b1fd067a65813f7f9ac5/F4.jpg", "media/b1fd067a65813f7f9ac5/F5.jpg", "media/b1fd067a65813f7f9ac5/F6.jpg", "media/b1fd067a65813f7f9ac5/F7.jpg", "media/b1fd067a65813f7f9ac5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in white shirt and purple shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. down\nD. up\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0ce787496671ffea3555b319b5fb828a0de5c0d5b2361de2cd385bc45a62210e"}
{"qa_id": "uvo__00HGQ6LqJZ8__camera_object_distance_change__01", "video_id": "uvo__00HGQ6LqJZ8", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, did object A (person in white shirt and purple shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b1fd067a65813f7f9ac5/F0.jpg", "media/b1fd067a65813f7f9ac5/F1.jpg", "media/b1fd067a65813f7f9ac5/F2.jpg", "media/b1fd067a65813f7f9ac5/F3.jpg", "media/b1fd067a65813f7f9ac5/F4.jpg", "media/b1fd067a65813f7f9ac5/F5.jpg", "media/b1fd067a65813f7f9ac5/F6.jpg", "media/b1fd067a65813f7f9ac5/F7.jpg", "media/b1fd067a65813f7f9ac5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person in white shirt and purple shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0077ab1642f749406615c455dac08706b7fbb1486c74d324f399dda552fc19dd"}
{"qa_id": "uvo__00Klr-1s1FY__camera_view_direction_change__00", "video_id": "uvo__00Klr-1s1FY", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/784dc1a2b8ee98a056dd/F0.jpg", "media/784dc1a2b8ee98a056dd/F1.jpg", "media/784dc1a2b8ee98a056dd/F2.jpg", "media/784dc1a2b8ee98a056dd/F3.jpg", "media/784dc1a2b8ee98a056dd/F4.jpg", "media/784dc1a2b8ee98a056dd/F5.jpg", "media/784dc1a2b8ee98a056dd/F6.jpg", "media/784dc1a2b8ee98a056dd/F7.jpg", "media/784dc1a2b8ee98a056dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "uvo__0347ZoDXyP0__camera_translation_direction__00", "video_id": "uvo__0347ZoDXyP0", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "backward", "up", "right"], "frame_paths": ["media/842f2ab7ca313e0bb046/F0.jpg", "media/842f2ab7ca313e0bb046/F1.jpg", "media/842f2ab7ca313e0bb046/F2.jpg", "media/842f2ab7ca313e0bb046/F3.jpg", "media/842f2ab7ca313e0bb046/F4.jpg", "media/842f2ab7ca313e0bb046/F5.jpg", "media/842f2ab7ca313e0bb046/F6.jpg", "media/842f2ab7ca313e0bb046/F7.jpg", "media/842f2ab7ca313e0bb046/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3d26b65c824c375ae1057a805e0a92f0d6dcfbccc8d4fed8871e7cd55d00eb62"}
{"qa_id": "uvo__035LtPeUFTE__camera_view_direction_change__00", "video_id": "uvo__035LtPeUFTE", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/a47fb6e5e3b97bf75155/F0.jpg", "media/a47fb6e5e3b97bf75155/F1.jpg", "media/a47fb6e5e3b97bf75155/F2.jpg", "media/a47fb6e5e3b97bf75155/F3.jpg", "media/a47fb6e5e3b97bf75155/F4.jpg", "media/a47fb6e5e3b97bf75155/F5.jpg", "media/a47fb6e5e3b97bf75155/F6.jpg", "media/a47fb6e5e3b97bf75155/F7.jpg", "media/a47fb6e5e3b97bf75155/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "603e61a89b8b9a39b97e3ae385c877f99d6aded5272823baf22c4e4d39c96f7d"}
{"qa_id": "uvo__03Ti8572JRY__camera_translation_direction__00", "video_id": "uvo__03Ti8572JRY", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "up", "right", "forward", "backward"], "frame_paths": ["media/a9505a118d145207f6fe/F0.jpg", "media/a9505a118d145207f6fe/F1.jpg", "media/a9505a118d145207f6fe/F2.jpg", "media/a9505a118d145207f6fe/F3.jpg", "media/a9505a118d145207f6fe/F4.jpg", "media/a9505a118d145207f6fe/F5.jpg", "media/a9505a118d145207f6fe/F6.jpg", "media/a9505a118d145207f6fe/F7.jpg", "media/a9505a118d145207f6fe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. up\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ae85c0966fc99e169819a557245490ddab9f7afea341ff60a64daff167e92f04"}
{"qa_id": "uvo__03Ti8572JRY__object_world_motion_direction__01", "video_id": "uvo__03Ti8572JRY", "task": "object_world_motion_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "forward", "right", "backward"], "frame_paths": ["media/4cc497ca389b75cd2f0c/F0.jpg", "media/4cc497ca389b75cd2f0c/F1.jpg", "media/4cc497ca389b75cd2f0c/F2.jpg", "media/4cc497ca389b75cd2f0c/F3.jpg", "media/4cc497ca389b75cd2f0c/F4.jpg", "media/4cc497ca389b75cd2f0c/F5.jpg", "media/4cc497ca389b75cd2f0c/F6.jpg", "media/4cc497ca389b75cd2f0c/F7.jpg", "media/4cc497ca389b75cd2f0c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e82f6aae87a7a71fb89c0d3b2df51710439bc09673690269b35d13360f0f6137"}
{"qa_id": "uvo__03Ti8572JRY__camera_object_distance_change__02", "video_id": "uvo__03Ti8572JRY", "task": "camera_object_distance_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, did object A (person in orange jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/4cc497ca389b75cd2f0c/F0.jpg", "media/4cc497ca389b75cd2f0c/F1.jpg", "media/4cc497ca389b75cd2f0c/F2.jpg", "media/4cc497ca389b75cd2f0c/F3.jpg", "media/4cc497ca389b75cd2f0c/F4.jpg", "media/4cc497ca389b75cd2f0c/F5.jpg", "media/4cc497ca389b75cd2f0c/F6.jpg", "media/4cc497ca389b75cd2f0c/F7.jpg", "media/4cc497ca389b75cd2f0c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in orange jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4638b447114d3d2181becd6d4cbed70ce422c5d2a77232f97f609450418a0c19"}
{"qa_id": "uvo__04yiI0AYPPQ__object_world_motion_direction__00", "video_id": "uvo__04yiI0AYPPQ", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in red shirt and orange pants) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "up", "forward", "down"], "frame_paths": ["media/d3ea60ac5a20ae2bd2e4/F0.jpg", "media/d3ea60ac5a20ae2bd2e4/F1.jpg", "media/d3ea60ac5a20ae2bd2e4/F2.jpg", "media/d3ea60ac5a20ae2bd2e4/F3.jpg", "media/d3ea60ac5a20ae2bd2e4/F4.jpg", "media/d3ea60ac5a20ae2bd2e4/F5.jpg", "media/d3ea60ac5a20ae2bd2e4/F6.jpg", "media/d3ea60ac5a20ae2bd2e4/F7.jpg", "media/d3ea60ac5a20ae2bd2e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in red shirt and orange pants) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "11823f58a7aed7df5275e6d009ee96f819ebfa2c4d5a8059ba79d18d0ce35251"}
{"qa_id": "uvo__05-73___szY__camera_object_distance_change__00", "video_id": "uvo__05-73___szY", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object B (person in dark blue jacket with white text on the back) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8b9bd11fd3a9b4ab87d5/F0.jpg", "media/8b9bd11fd3a9b4ab87d5/F1.jpg", "media/8b9bd11fd3a9b4ab87d5/F2.jpg", "media/8b9bd11fd3a9b4ab87d5/F3.jpg", "media/8b9bd11fd3a9b4ab87d5/F4.jpg", "media/8b9bd11fd3a9b4ab87d5/F5.jpg", "media/8b9bd11fd3a9b4ab87d5/F6.jpg", "media/8b9bd11fd3a9b4ab87d5/F7.jpg", "media/8b9bd11fd3a9b4ab87d5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (person in dark blue jacket with white text on the back) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "76f15ed5e3a2296fef2c2ee6c6c376f72e7ebc2986c95c2aa3c2ae6d3f45444e"}
{"qa_id": "uvo__05-73___szY__object_pair_distance_change__01", "video_id": "uvo__05-73___szY", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (person in red jacket standing at a microphone) and object B (person in dark blue jacket with white text on the back) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/bf693298eb958abf9853/F0.jpg", "media/bf693298eb958abf9853/F1.jpg", "media/bf693298eb958abf9853/F2.jpg", "media/bf693298eb958abf9853/F3.jpg", "media/bf693298eb958abf9853/F4.jpg", "media/bf693298eb958abf9853/F5.jpg", "media/bf693298eb958abf9853/F6.jpg", "media/bf693298eb958abf9853/F7.jpg", "media/bf693298eb958abf9853/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in red jacket standing at a microphone) and object B (person in dark blue jacket with white text on the back) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "902578dc613aca19bf6bce0fe217c1264e31c00ac69e82028c3a10e2592c8548"}
{"qa_id": "uvo__06DYNQ5z00E__object_world_motion_direction__00", "video_id": "uvo__06DYNQ5z00E", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (man in dark t-shirt and shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "backward", "up", "forward"], "frame_paths": ["media/93fb33e6e9bd2113db48/F0.jpg", "media/93fb33e6e9bd2113db48/F1.jpg", "media/93fb33e6e9bd2113db48/F2.jpg", "media/93fb33e6e9bd2113db48/F3.jpg", "media/93fb33e6e9bd2113db48/F4.jpg", "media/93fb33e6e9bd2113db48/F5.jpg", "media/93fb33e6e9bd2113db48/F6.jpg", "media/93fb33e6e9bd2113db48/F7.jpg", "media/93fb33e6e9bd2113db48/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (man in dark t-shirt and shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee9b7213bf318593d1e2518e01479a040ffbb65757c62208061c7ac4126b6aae"}
{"qa_id": "uvo__06ZqVRSZ8wE__camera_view_direction_change__00", "video_id": "uvo__06ZqVRSZ8wE", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/fa43677f8473fb97d42a/F0.jpg", "media/fa43677f8473fb97d42a/F1.jpg", "media/fa43677f8473fb97d42a/F2.jpg", "media/fa43677f8473fb97d42a/F3.jpg", "media/fa43677f8473fb97d42a/F4.jpg", "media/fa43677f8473fb97d42a/F5.jpg", "media/fa43677f8473fb97d42a/F6.jpg", "media/fa43677f8473fb97d42a/F7.jpg", "media/fa43677f8473fb97d42a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1b8dee0cddab4122077885c2e5c7494e89b36840368247757b40530a4196017"}
{"qa_id": "uvo__06kWYuf1qrU__camera_translation_direction__00", "video_id": "uvo__06kWYuf1qrU", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "backward", "left", "down"], "frame_paths": ["media/dba310cfb5800e080beb/F0.jpg", "media/dba310cfb5800e080beb/F1.jpg", "media/dba310cfb5800e080beb/F2.jpg", "media/dba310cfb5800e080beb/F3.jpg", "media/dba310cfb5800e080beb/F4.jpg", "media/dba310cfb5800e080beb/F5.jpg", "media/dba310cfb5800e080beb/F6.jpg", "media/dba310cfb5800e080beb/F7.jpg", "media/dba310cfb5800e080beb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa1373235775277ddcf470dcc35e97ab172ea1533fa310d98c3da118b2ad73f4"}
{"qa_id": "uvo__06kWYuf1qrU__camera_view_direction_change__01", "video_id": "uvo__06kWYuf1qrU", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/dba310cfb5800e080beb/F0.jpg", "media/dba310cfb5800e080beb/F1.jpg", "media/dba310cfb5800e080beb/F2.jpg", "media/dba310cfb5800e080beb/F3.jpg", "media/dba310cfb5800e080beb/F4.jpg", "media/dba310cfb5800e080beb/F5.jpg", "media/dba310cfb5800e080beb/F6.jpg", "media/dba310cfb5800e080beb/F7.jpg", "media/dba310cfb5800e080beb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "uvo__07j7tJxpe5M__object_world_motion_direction__00", "video_id": "uvo__07j7tJxpe5M", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (shirtless person in dark shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "right", "backward", "up", "left"], "frame_paths": ["media/a46fa6755973f54c1ce0/F0.jpg", "media/a46fa6755973f54c1ce0/F1.jpg", "media/a46fa6755973f54c1ce0/F2.jpg", "media/a46fa6755973f54c1ce0/F3.jpg", "media/a46fa6755973f54c1ce0/F4.jpg", "media/a46fa6755973f54c1ce0/F5.jpg", "media/a46fa6755973f54c1ce0/F6.jpg", "media/a46fa6755973f54c1ce0/F7.jpg", "media/a46fa6755973f54c1ce0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (shirtless person in dark shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c7f06552d9d95117ca218428fb90400e00203701ce05cb8187b4b46881ac4223"}
{"qa_id": "uvo__07j7tJxpe5M__camera_object_distance_change__01", "video_id": "uvo__07j7tJxpe5M", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object A (shirtless person in dark shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a46fa6755973f54c1ce0/F0.jpg", "media/a46fa6755973f54c1ce0/F1.jpg", "media/a46fa6755973f54c1ce0/F2.jpg", "media/a46fa6755973f54c1ce0/F3.jpg", "media/a46fa6755973f54c1ce0/F4.jpg", "media/a46fa6755973f54c1ce0/F5.jpg", "media/a46fa6755973f54c1ce0/F6.jpg", "media/a46fa6755973f54c1ce0/F7.jpg", "media/a46fa6755973f54c1ce0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (shirtless person in dark shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ff5c7fcf1944d501203d9af61cfb225442be5b616c098221af9eee19a9db511c"}
{"qa_id": "uvo__081YeOmMfRg__object_world_motion_direction__00", "video_id": "uvo__081YeOmMfRg", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark jacket cross-country skiing) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "forward", "right", "up", "left"], "frame_paths": ["media/d0e5f3f1e81ce781f96d/F0.jpg", "media/d0e5f3f1e81ce781f96d/F1.jpg", "media/d0e5f3f1e81ce781f96d/F2.jpg", "media/d0e5f3f1e81ce781f96d/F3.jpg", "media/d0e5f3f1e81ce781f96d/F4.jpg", "media/d0e5f3f1e81ce781f96d/F5.jpg", "media/d0e5f3f1e81ce781f96d/F6.jpg", "media/d0e5f3f1e81ce781f96d/F7.jpg", "media/d0e5f3f1e81ce781f96d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark jacket cross-country skiing) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. forward\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3827a0c79ea69a05fcf781890af238a1b520aa7fe5148bda6cab5db6393e2889"}
{"qa_id": "uvo__081YeOmMfRg__camera_object_distance_change__01", "video_id": "uvo__081YeOmMfRg", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object A (person in dark jacket cross-country skiing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d0e5f3f1e81ce781f96d/F0.jpg", "media/d0e5f3f1e81ce781f96d/F1.jpg", "media/d0e5f3f1e81ce781f96d/F2.jpg", "media/d0e5f3f1e81ce781f96d/F3.jpg", "media/d0e5f3f1e81ce781f96d/F4.jpg", "media/d0e5f3f1e81ce781f96d/F5.jpg", "media/d0e5f3f1e81ce781f96d/F6.jpg", "media/d0e5f3f1e81ce781f96d/F7.jpg", "media/d0e5f3f1e81ce781f96d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (person in dark jacket cross-country skiing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "89293146ff698e3690450f76c16bf4b5f3a9fa866c8e19f0de6ca1e8d200b2fe"}
{"qa_id": "uvo__09sCD17AOyY__camera_view_direction_change__00", "video_id": "uvo__09sCD17AOyY", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/b5df74b34063d9b07eaa/F0.jpg", "media/b5df74b34063d9b07eaa/F1.jpg", "media/b5df74b34063d9b07eaa/F2.jpg", "media/b5df74b34063d9b07eaa/F3.jpg", "media/b5df74b34063d9b07eaa/F4.jpg", "media/b5df74b34063d9b07eaa/F5.jpg", "media/b5df74b34063d9b07eaa/F6.jpg", "media/b5df74b34063d9b07eaa/F7.jpg", "media/b5df74b34063d9b07eaa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "uvo__0A61TZLTGGg__camera_view_direction_change__00", "video_id": "uvo__0A61TZLTGGg", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/79c6c25398e9ffb89903/F0.jpg", "media/79c6c25398e9ffb89903/F1.jpg", "media/79c6c25398e9ffb89903/F2.jpg", "media/79c6c25398e9ffb89903/F3.jpg", "media/79c6c25398e9ffb89903/F4.jpg", "media/79c6c25398e9ffb89903/F5.jpg", "media/79c6c25398e9ffb89903/F6.jpg", "media/79c6c25398e9ffb89903/F7.jpg", "media/79c6c25398e9ffb89903/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8bdeec091b4f2417ac9b70bcffa998a9a1405e7877dbb03cea4c3bda53dfb11e"}
{"qa_id": "uvo__0ATPzLSCCTg__camera_translation_direction__00", "video_id": "uvo__0ATPzLSCCTg", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/cc84182b5ae5e68896f8/F0.jpg", "media/cc84182b5ae5e68896f8/F1.jpg", "media/cc84182b5ae5e68896f8/F2.jpg", "media/cc84182b5ae5e68896f8/F3.jpg", "media/cc84182b5ae5e68896f8/F4.jpg", "media/cc84182b5ae5e68896f8/F5.jpg", "media/cc84182b5ae5e68896f8/F6.jpg", "media/cc84182b5ae5e68896f8/F7.jpg", "media/cc84182b5ae5e68896f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f86a65f6d402a478c10f31d5f3ef698b86752829f67e51ceb817f40da7dd5711"}
{"qa_id": "uvo__0ATPzLSCCTg__camera_view_direction_change__01", "video_id": "uvo__0ATPzLSCCTg", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/cc84182b5ae5e68896f8/F0.jpg", "media/cc84182b5ae5e68896f8/F1.jpg", "media/cc84182b5ae5e68896f8/F2.jpg", "media/cc84182b5ae5e68896f8/F3.jpg", "media/cc84182b5ae5e68896f8/F4.jpg", "media/cc84182b5ae5e68896f8/F5.jpg", "media/cc84182b5ae5e68896f8/F6.jpg", "media/cc84182b5ae5e68896f8/F7.jpg", "media/cc84182b5ae5e68896f8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654016ae105affb53bc2face594c67e96f587f4cdc806a6f418076b30032d255"}
{"qa_id": "uvo__0ApEeAZnMqk__object_world_motion_direction__00", "video_id": "uvo__0ApEeAZnMqk", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "left", "forward"], "frame_paths": ["media/55030ab47b4967458ddd/F0.jpg", "media/55030ab47b4967458ddd/F1.jpg", "media/55030ab47b4967458ddd/F2.jpg", "media/55030ab47b4967458ddd/F3.jpg", "media/55030ab47b4967458ddd/F4.jpg", "media/55030ab47b4967458ddd/F5.jpg", "media/55030ab47b4967458ddd/F6.jpg", "media/55030ab47b4967458ddd/F7.jpg", "media/55030ab47b4967458ddd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in orange jacket) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f162b09fca1934c597844d1077e3ce30e64e0fc70f7f8a49caa047c199f390e5"}
{"qa_id": "uvo__0BLyz0GTX8w__camera_translation_direction__00", "video_id": "uvo__0BLyz0GTX8w", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "forward", "up", "down"], "frame_paths": ["media/43cddfc63b1ac71ce277/F0.jpg", "media/43cddfc63b1ac71ce277/F1.jpg", "media/43cddfc63b1ac71ce277/F2.jpg", "media/43cddfc63b1ac71ce277/F3.jpg", "media/43cddfc63b1ac71ce277/F4.jpg", "media/43cddfc63b1ac71ce277/F5.jpg", "media/43cddfc63b1ac71ce277/F6.jpg", "media/43cddfc63b1ac71ce277/F7.jpg", "media/43cddfc63b1ac71ce277/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2847a3b7a033b73f261d400ee2874380fb2bd0940831d75d8c090b543c982b68"}
{"qa_id": "uvo__0Bs9xxi8sxY__camera_view_direction_change__00", "video_id": "uvo__0Bs9xxi8sxY", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/0d1b85d62722eea4dce5/F0.jpg", "media/0d1b85d62722eea4dce5/F1.jpg", "media/0d1b85d62722eea4dce5/F2.jpg", "media/0d1b85d62722eea4dce5/F3.jpg", "media/0d1b85d62722eea4dce5/F4.jpg", "media/0d1b85d62722eea4dce5/F5.jpg", "media/0d1b85d62722eea4dce5/F6.jpg", "media/0d1b85d62722eea4dce5/F7.jpg", "media/0d1b85d62722eea4dce5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a74873eaad6f5b5bc7fa11ba8a6c28fb60d68336f8fd3430e31ab7a810d2526d"}
{"qa_id": "uvo__0Bs9xxi8sxY__object_world_motion_direction__01", "video_id": "uvo__0Bs9xxi8sxY", "task": "object_world_motion_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in black shirt and white shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "right", "down"], "frame_paths": ["media/6bd9d15a18d7ba963d2e/F0.jpg", "media/6bd9d15a18d7ba963d2e/F1.jpg", "media/6bd9d15a18d7ba963d2e/F2.jpg", "media/6bd9d15a18d7ba963d2e/F3.jpg", "media/6bd9d15a18d7ba963d2e/F4.jpg", "media/6bd9d15a18d7ba963d2e/F5.jpg", "media/6bd9d15a18d7ba963d2e/F6.jpg", "media/6bd9d15a18d7ba963d2e/F7.jpg", "media/6bd9d15a18d7ba963d2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in black shirt and white shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "667b5768c72697136e6a87c2bb8f9c33a020f6669b22110ca75ad384c144ae19"}
{"qa_id": "uvo__0Bs9xxi8sxY__camera_object_distance_change__02", "video_id": "uvo__0Bs9xxi8sxY", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F5, did object A (person in black shirt and white shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/6bd9d15a18d7ba963d2e/F0.jpg", "media/6bd9d15a18d7ba963d2e/F1.jpg", "media/6bd9d15a18d7ba963d2e/F2.jpg", "media/6bd9d15a18d7ba963d2e/F3.jpg", "media/6bd9d15a18d7ba963d2e/F4.jpg", "media/6bd9d15a18d7ba963d2e/F5.jpg", "media/6bd9d15a18d7ba963d2e/F6.jpg", "media/6bd9d15a18d7ba963d2e/F7.jpg", "media/6bd9d15a18d7ba963d2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did object A (person in black shirt and white shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7d94539d9d1565b10a2a81eb8daee143ee111be12b6b5449b652434320a1ae0a"}
{"qa_id": "uvo__0C3barg2QKI__camera_view_direction_change__00", "video_id": "uvo__0C3barg2QKI", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/336d2c23ef568b837fe1/F0.jpg", "media/336d2c23ef568b837fe1/F1.jpg", "media/336d2c23ef568b837fe1/F2.jpg", "media/336d2c23ef568b837fe1/F3.jpg", "media/336d2c23ef568b837fe1/F4.jpg", "media/336d2c23ef568b837fe1/F5.jpg", "media/336d2c23ef568b837fe1/F6.jpg", "media/336d2c23ef568b837fe1/F7.jpg", "media/336d2c23ef568b837fe1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88569cd745de284a24111d02b2fdec72688e7586531e7f8356c77faa9dd1fd28"}
{"qa_id": "uvo__0DdXhnmSEbE__camera_translation_direction__00", "video_id": "uvo__0DdXhnmSEbE", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "left", "right", "down", "up"], "frame_paths": ["media/201aebb96210e1a7bf0e/F0.jpg", "media/201aebb96210e1a7bf0e/F1.jpg", "media/201aebb96210e1a7bf0e/F2.jpg", "media/201aebb96210e1a7bf0e/F3.jpg", "media/201aebb96210e1a7bf0e/F4.jpg", "media/201aebb96210e1a7bf0e/F5.jpg", "media/201aebb96210e1a7bf0e/F6.jpg", "media/201aebb96210e1a7bf0e/F7.jpg", "media/201aebb96210e1a7bf0e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. left\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c24a7d342232a75f8c78ab8a07c7efda94f26e177672cf07f2f65a245710d23"}
{"qa_id": "uvo__0DfXV_B7WvI__camera_view_direction_change__00", "video_id": "uvo__0DfXV_B7WvI", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/d66e6bc53abd993dea3d/F0.jpg", "media/d66e6bc53abd993dea3d/F1.jpg", "media/d66e6bc53abd993dea3d/F2.jpg", "media/d66e6bc53abd993dea3d/F3.jpg", "media/d66e6bc53abd993dea3d/F4.jpg", "media/d66e6bc53abd993dea3d/F5.jpg", "media/d66e6bc53abd993dea3d/F6.jpg", "media/d66e6bc53abd993dea3d/F7.jpg", "media/d66e6bc53abd993dea3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f9558cfcb39b4b8062b0685a5042b148145270b4c4b58790fc901bc14fc2ec8"}
{"qa_id": "uvo__0E299kpi2ow__camera_translation_direction__00", "video_id": "uvo__0E299kpi2ow", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "up", "down", "backward"], "frame_paths": ["media/cf6d67616470a674b2ec/F0.jpg", "media/cf6d67616470a674b2ec/F1.jpg", "media/cf6d67616470a674b2ec/F2.jpg", "media/cf6d67616470a674b2ec/F3.jpg", "media/cf6d67616470a674b2ec/F4.jpg", "media/cf6d67616470a674b2ec/F5.jpg", "media/cf6d67616470a674b2ec/F6.jpg", "media/cf6d67616470a674b2ec/F7.jpg", "media/cf6d67616470a674b2ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e09bc893385398489563b11c96ff0076ec358243e34510775cba9c8ee5d9bac3"}
{"qa_id": "uvo__0Em6soMELS0__object_world_motion_direction__00", "video_id": "uvo__0Em6soMELS0", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (person in a white t-shirt and dark pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "down", "up", "left"], "frame_paths": ["media/09db5bbf2f36282dad91/F0.jpg", "media/09db5bbf2f36282dad91/F1.jpg", "media/09db5bbf2f36282dad91/F2.jpg", "media/09db5bbf2f36282dad91/F3.jpg", "media/09db5bbf2f36282dad91/F4.jpg", "media/09db5bbf2f36282dad91/F5.jpg", "media/09db5bbf2f36282dad91/F6.jpg", "media/09db5bbf2f36282dad91/F7.jpg", "media/09db5bbf2f36282dad91/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (person in a white t-shirt and dark pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5a5480afc2996ef11ce0ca9925955b4091664f45b699ecef4d1e39593c6bb97e"}
{"qa_id": "uvo__0Em6soMELS0__object_pair_distance_change__01", "video_id": "uvo__0Em6soMELS0", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F6, did the 3D distance between object A (person in dark clothing sitting on a ledge) and object B (person in a white t-shirt and dark pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/201fbd4b27fac464be5a/F0.jpg", "media/201fbd4b27fac464be5a/F1.jpg", "media/201fbd4b27fac464be5a/F2.jpg", "media/201fbd4b27fac464be5a/F3.jpg", "media/201fbd4b27fac464be5a/F4.jpg", "media/201fbd4b27fac464be5a/F5.jpg", "media/201fbd4b27fac464be5a/F6.jpg", "media/201fbd4b27fac464be5a/F7.jpg", "media/201fbd4b27fac464be5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (person in dark clothing sitting on a ledge) and object B (person in a white t-shirt and dark pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cf3ff5efdaa53f489d5bc53ac05dace7868c445d10c09bf5a55c92e6a073c8e1"}
{"qa_id": "uvo__0F1WmMXjml4__camera_view_direction_change__00", "video_id": "uvo__0F1WmMXjml4", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/1f945a77e5dd28e049b5/F0.jpg", "media/1f945a77e5dd28e049b5/F1.jpg", "media/1f945a77e5dd28e049b5/F2.jpg", "media/1f945a77e5dd28e049b5/F3.jpg", "media/1f945a77e5dd28e049b5/F4.jpg", "media/1f945a77e5dd28e049b5/F5.jpg", "media/1f945a77e5dd28e049b5/F6.jpg", "media/1f945a77e5dd28e049b5/F7.jpg", "media/1f945a77e5dd28e049b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__0GUtW9-GQQw__camera_object_distance_change__00", "video_id": "uvo__0GUtW9-GQQw", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, did object A (boy in blue shirt and dark shorts running) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a02d015e9e1d4d7bf229/F0.jpg", "media/a02d015e9e1d4d7bf229/F1.jpg", "media/a02d015e9e1d4d7bf229/F2.jpg", "media/a02d015e9e1d4d7bf229/F3.jpg", "media/a02d015e9e1d4d7bf229/F4.jpg", "media/a02d015e9e1d4d7bf229/F5.jpg", "media/a02d015e9e1d4d7bf229/F6.jpg", "media/a02d015e9e1d4d7bf229/F7.jpg", "media/a02d015e9e1d4d7bf229/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (boy in blue shirt and dark shorts running) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "90de51f59e746f8bc91208610919176274663ffdf3a36a41bc4df366fe1cfbc7"}
{"qa_id": "uvo__0GpOe8tgrRo__camera_translation_direction__00", "video_id": "uvo__0GpOe8tgrRo", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "up", "forward", "right"], "frame_paths": ["media/4a919579d002767a961d/F0.jpg", "media/4a919579d002767a961d/F1.jpg", "media/4a919579d002767a961d/F2.jpg", "media/4a919579d002767a961d/F3.jpg", "media/4a919579d002767a961d/F4.jpg", "media/4a919579d002767a961d/F5.jpg", "media/4a919579d002767a961d/F6.jpg", "media/4a919579d002767a961d/F7.jpg", "media/4a919579d002767a961d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "139777e4267bb8e1b5ec040d500c191f669409c04a18b6a6946181880521b9fa"}
{"qa_id": "uvo__0GpOe8tgrRo__camera_view_direction_change__01", "video_id": "uvo__0GpOe8tgrRo", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/4a919579d002767a961d/F0.jpg", "media/4a919579d002767a961d/F1.jpg", "media/4a919579d002767a961d/F2.jpg", "media/4a919579d002767a961d/F3.jpg", "media/4a919579d002767a961d/F4.jpg", "media/4a919579d002767a961d/F5.jpg", "media/4a919579d002767a961d/F6.jpg", "media/4a919579d002767a961d/F7.jpg", "media/4a919579d002767a961d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a4b2c38de6a7b4852267bb81e22d1c3ca88491c502534af149743550da5fde88"}
{"qa_id": "uvo__0H1C-nllaWk__camera_view_direction_change__00", "video_id": "uvo__0H1C-nllaWk", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/df88f0de00e13b2a3e5b/F0.jpg", "media/df88f0de00e13b2a3e5b/F1.jpg", "media/df88f0de00e13b2a3e5b/F2.jpg", "media/df88f0de00e13b2a3e5b/F3.jpg", "media/df88f0de00e13b2a3e5b/F4.jpg", "media/df88f0de00e13b2a3e5b/F5.jpg", "media/df88f0de00e13b2a3e5b/F6.jpg", "media/df88f0de00e13b2a3e5b/F7.jpg", "media/df88f0de00e13b2a3e5b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "uvo__0Hv8j33iHpU__camera_translation_direction__00", "video_id": "uvo__0Hv8j33iHpU", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "left", "down", "backward", "right"], "frame_paths": ["media/e9818e77424285483187/F0.jpg", "media/e9818e77424285483187/F1.jpg", "media/e9818e77424285483187/F2.jpg", "media/e9818e77424285483187/F3.jpg", "media/e9818e77424285483187/F4.jpg", "media/e9818e77424285483187/F5.jpg", "media/e9818e77424285483187/F6.jpg", "media/e9818e77424285483187/F7.jpg", "media/e9818e77424285483187/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. left\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab3f5b3078aa3f3cc28f09d66536a8f886d237b16dc2a7204c2c2841942b4ee4"}
{"qa_id": "uvo__0Hv8j33iHpU__object_world_motion_direction__01", "video_id": "uvo__0Hv8j33iHpU", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object A (person in white top and orange pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "right", "left", "forward"], "frame_paths": ["media/2c5a582dee23ad317672/F0.jpg", "media/2c5a582dee23ad317672/F1.jpg", "media/2c5a582dee23ad317672/F2.jpg", "media/2c5a582dee23ad317672/F3.jpg", "media/2c5a582dee23ad317672/F4.jpg", "media/2c5a582dee23ad317672/F5.jpg", "media/2c5a582dee23ad317672/F6.jpg", "media/2c5a582dee23ad317672/F7.jpg", "media/2c5a582dee23ad317672/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object A (person in white top and orange pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6bf2a6e353b8754ba9056db43ca86d953d0f8da7605b45bbf0a5822745ce7d88"}
{"qa_id": "uvo__0Hv8j33iHpU__camera_object_distance_change__02", "video_id": "uvo__0Hv8j33iHpU", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, did object A (person in white top and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2c5a582dee23ad317672/F0.jpg", "media/2c5a582dee23ad317672/F1.jpg", "media/2c5a582dee23ad317672/F2.jpg", "media/2c5a582dee23ad317672/F3.jpg", "media/2c5a582dee23ad317672/F4.jpg", "media/2c5a582dee23ad317672/F5.jpg", "media/2c5a582dee23ad317672/F6.jpg", "media/2c5a582dee23ad317672/F7.jpg", "media/2c5a582dee23ad317672/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (person in white top and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87571eed97c5ef66e4fe3fb6526372f24e3f91ae4c97a26295c4123da5c206a5"}
{"qa_id": "uvo__0Hzl3c8aUXI__object_world_motion_direction__00", "video_id": "uvo__0Hzl3c8aUXI", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in dark clothing standing on a grassy field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "forward", "up", "down"], "frame_paths": ["media/69f9d1d921a7d5e1ded2/F0.jpg", "media/69f9d1d921a7d5e1ded2/F1.jpg", "media/69f9d1d921a7d5e1ded2/F2.jpg", "media/69f9d1d921a7d5e1ded2/F3.jpg", "media/69f9d1d921a7d5e1ded2/F4.jpg", "media/69f9d1d921a7d5e1ded2/F5.jpg", "media/69f9d1d921a7d5e1ded2/F6.jpg", "media/69f9d1d921a7d5e1ded2/F7.jpg", "media/69f9d1d921a7d5e1ded2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in dark clothing standing on a grassy field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f759090b2d9315dbb195c79254dd1baf21880f7d0c7a03e0468276c8db3c9064"}
{"qa_id": "uvo__0JZm3c-ZhKo__object_travel_distance_comparison__00", "video_id": "uvo__0JZm3c-ZhKo", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and dark pants on a skateboard) or B (small child in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/4c87a25ec74133eb74b2/F0.jpg", "media/4c87a25ec74133eb74b2/F1.jpg", "media/4c87a25ec74133eb74b2/F2.jpg", "media/4c87a25ec74133eb74b2/F3.jpg", "media/4c87a25ec74133eb74b2/F4.jpg", "media/4c87a25ec74133eb74b2/F5.jpg", "media/4c87a25ec74133eb74b2/F6.jpg", "media/4c87a25ec74133eb74b2/F7.jpg", "media/4c87a25ec74133eb74b2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in blue shirt and dark pants on a skateboard) or B (small child in dark clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d27f851fe57ace6f9f1b1b07212e693730855d8b1f6f44f993a2eab0089c779"}
{"qa_id": "uvo__0KNJtZdmFPI__camera_translation_direction__00", "video_id": "uvo__0KNJtZdmFPI", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "up", "backward", "forward"], "frame_paths": ["media/37b6cd45a943d5cd00d3/F0.jpg", "media/37b6cd45a943d5cd00d3/F1.jpg", "media/37b6cd45a943d5cd00d3/F2.jpg", "media/37b6cd45a943d5cd00d3/F3.jpg", "media/37b6cd45a943d5cd00d3/F4.jpg", "media/37b6cd45a943d5cd00d3/F5.jpg", "media/37b6cd45a943d5cd00d3/F6.jpg", "media/37b6cd45a943d5cd00d3/F7.jpg", "media/37b6cd45a943d5cd00d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "80a7bf91f7af8a0e62718eade6aaad817fe16e842b5da26d88db241dd0b660bb"}
{"qa_id": "uvo__0L3RDU3j3h4__camera_translation_direction__00", "video_id": "uvo__0L3RDU3j3h4", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "up", "down", "backward", "left"], "frame_paths": ["media/7a9f39a6812d6b18ada7/F0.jpg", "media/7a9f39a6812d6b18ada7/F1.jpg", "media/7a9f39a6812d6b18ada7/F2.jpg", "media/7a9f39a6812d6b18ada7/F3.jpg", "media/7a9f39a6812d6b18ada7/F4.jpg", "media/7a9f39a6812d6b18ada7/F5.jpg", "media/7a9f39a6812d6b18ada7/F6.jpg", "media/7a9f39a6812d6b18ada7/F7.jpg", "media/7a9f39a6812d6b18ada7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. up\nD. down\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "700daf3e7453886803c403ab3929eddfef2b56c14366c369cd1757a11feae101"}
{"qa_id": "uvo__0L3RDU3j3h4__camera_view_direction_change__01", "video_id": "uvo__0L3RDU3j3h4", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/7a9f39a6812d6b18ada7/F0.jpg", "media/7a9f39a6812d6b18ada7/F1.jpg", "media/7a9f39a6812d6b18ada7/F2.jpg", "media/7a9f39a6812d6b18ada7/F3.jpg", "media/7a9f39a6812d6b18ada7/F4.jpg", "media/7a9f39a6812d6b18ada7/F5.jpg", "media/7a9f39a6812d6b18ada7/F6.jpg", "media/7a9f39a6812d6b18ada7/F7.jpg", "media/7a9f39a6812d6b18ada7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4be0070f398ed05c208baca3c0bbabd077ee5b3b77ffd593732c4fb964889062"}
{"qa_id": "uvo__0L4hQd-kGIQ__camera_view_direction_change__00", "video_id": "uvo__0L4hQd-kGIQ", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/56ef01789b3a394c4dd0/F0.jpg", "media/56ef01789b3a394c4dd0/F1.jpg", "media/56ef01789b3a394c4dd0/F2.jpg", "media/56ef01789b3a394c4dd0/F3.jpg", "media/56ef01789b3a394c4dd0/F4.jpg", "media/56ef01789b3a394c4dd0/F5.jpg", "media/56ef01789b3a394c4dd0/F6.jpg", "media/56ef01789b3a394c4dd0/F7.jpg", "media/56ef01789b3a394c4dd0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2caa625aa254fd3abebce32193d4f99365c506ed6ff0451a16628a0cfc0281cc"}
{"qa_id": "uvo__0LFfu9I99VI__object_pair_distance_change__00", "video_id": "uvo__0LFfu9I99VI", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, did the 3D distance between object A (person in black top and shorts standing on a red segway) and object B (person in dark clothing riding a bicycle) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/6fe9bd5ef259ce2b6c52/F0.jpg", "media/6fe9bd5ef259ce2b6c52/F1.jpg", "media/6fe9bd5ef259ce2b6c52/F2.jpg", "media/6fe9bd5ef259ce2b6c52/F3.jpg", "media/6fe9bd5ef259ce2b6c52/F4.jpg", "media/6fe9bd5ef259ce2b6c52/F5.jpg", "media/6fe9bd5ef259ce2b6c52/F6.jpg", "media/6fe9bd5ef259ce2b6c52/F7.jpg", "media/6fe9bd5ef259ce2b6c52/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (person in black top and shorts standing on a red segway) and object B (person in dark clothing riding a bicycle) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b454fd9dab112122c1ded91cc0038b029f6c08225a63df6f76237f45894666c"}
{"qa_id": "uvo__0LOJzsBnOpI__object_world_motion_direction__00", "video_id": "uvo__0LOJzsBnOpI", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (person in dark clothing standing in a snowy field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "down", "up", "left"], "frame_paths": ["media/85ceb6a1e672d2aa40da/F0.jpg", "media/85ceb6a1e672d2aa40da/F1.jpg", "media/85ceb6a1e672d2aa40da/F2.jpg", "media/85ceb6a1e672d2aa40da/F3.jpg", "media/85ceb6a1e672d2aa40da/F4.jpg", "media/85ceb6a1e672d2aa40da/F5.jpg", "media/85ceb6a1e672d2aa40da/F6.jpg", "media/85ceb6a1e672d2aa40da/F7.jpg", "media/85ceb6a1e672d2aa40da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (person in dark clothing standing in a snowy field) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "97cb3d915d24c2f1214758c8c7b5f755040ecbef5793deb10ec61f46a539a653"}
{"qa_id": "uvo__0MVxFYLQbic__camera_view_direction_change__00", "video_id": "uvo__0MVxFYLQbic", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/43398acbfabc94a4b702/F0.jpg", "media/43398acbfabc94a4b702/F1.jpg", "media/43398acbfabc94a4b702/F2.jpg", "media/43398acbfabc94a4b702/F3.jpg", "media/43398acbfabc94a4b702/F4.jpg", "media/43398acbfabc94a4b702/F5.jpg", "media/43398acbfabc94a4b702/F6.jpg", "media/43398acbfabc94a4b702/F7.jpg", "media/43398acbfabc94a4b702/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "uvo__0N1odE53B0g__camera_view_direction_change__00", "video_id": "uvo__0N1odE53B0g", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/2a28026c8014143bc88f/F0.jpg", "media/2a28026c8014143bc88f/F1.jpg", "media/2a28026c8014143bc88f/F2.jpg", "media/2a28026c8014143bc88f/F3.jpg", "media/2a28026c8014143bc88f/F4.jpg", "media/2a28026c8014143bc88f/F5.jpg", "media/2a28026c8014143bc88f/F6.jpg", "media/2a28026c8014143bc88f/F7.jpg", "media/2a28026c8014143bc88f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1c42597cfebf6755eabac9c1dfd855f299f0d508ac78d39e51d22ad83c9dd15e"}
{"qa_id": "uvo__0NDp0DgOHxs__camera_translation_direction__00", "video_id": "uvo__0NDp0DgOHxs", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "backward", "forward", "left"], "frame_paths": ["media/79aec22c0c6e05b590f6/F0.jpg", "media/79aec22c0c6e05b590f6/F1.jpg", "media/79aec22c0c6e05b590f6/F2.jpg", "media/79aec22c0c6e05b590f6/F3.jpg", "media/79aec22c0c6e05b590f6/F4.jpg", "media/79aec22c0c6e05b590f6/F5.jpg", "media/79aec22c0c6e05b590f6/F6.jpg", "media/79aec22c0c6e05b590f6/F7.jpg", "media/79aec22c0c6e05b590f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a745d54d004b1318c0372a80beb35f6f07026e94d97e0aff8d2fb6b7ba34e9fd"}
{"qa_id": "uvo__0NUTcXmBjRs__camera_translation_direction__00", "video_id": "uvo__0NUTcXmBjRs", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "up", "forward", "down"], "frame_paths": ["media/fd5494bd359b8a59476a/F0.jpg", "media/fd5494bd359b8a59476a/F1.jpg", "media/fd5494bd359b8a59476a/F2.jpg", "media/fd5494bd359b8a59476a/F3.jpg", "media/fd5494bd359b8a59476a/F4.jpg", "media/fd5494bd359b8a59476a/F5.jpg", "media/fd5494bd359b8a59476a/F6.jpg", "media/fd5494bd359b8a59476a/F7.jpg", "media/fd5494bd359b8a59476a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. up\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "655a80f0425de1275f451a3f0965e3221f44b06975e4c972e3e36f629c2332e5"}
{"qa_id": "uvo__0NUTcXmBjRs__camera_view_direction_change__01", "video_id": "uvo__0NUTcXmBjRs", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/fd5494bd359b8a59476a/F0.jpg", "media/fd5494bd359b8a59476a/F1.jpg", "media/fd5494bd359b8a59476a/F2.jpg", "media/fd5494bd359b8a59476a/F3.jpg", "media/fd5494bd359b8a59476a/F4.jpg", "media/fd5494bd359b8a59476a/F5.jpg", "media/fd5494bd359b8a59476a/F6.jpg", "media/fd5494bd359b8a59476a/F7.jpg", "media/fd5494bd359b8a59476a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a234d67b737788407d244165dc03a879ad573d5b8648751041e9b4b6172252fb"}
{"qa_id": "uvo__0NUTcXmBjRs__object_world_motion_direction__02", "video_id": "uvo__0NUTcXmBjRs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (boy in blue shirt riding a green toy tractor) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "forward", "up"], "frame_paths": ["media/be15fd0ea384e9856284/F0.jpg", "media/be15fd0ea384e9856284/F1.jpg", "media/be15fd0ea384e9856284/F2.jpg", "media/be15fd0ea384e9856284/F3.jpg", "media/be15fd0ea384e9856284/F4.jpg", "media/be15fd0ea384e9856284/F5.jpg", "media/be15fd0ea384e9856284/F6.jpg", "media/be15fd0ea384e9856284/F7.jpg", "media/be15fd0ea384e9856284/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (boy in blue shirt riding a green toy tractor) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88f45dd1eb154730b5ebf71beef486b710a36185bda6215e6746f80d8872c592"}
{"qa_id": "uvo__0NUTcXmBjRs__camera_object_distance_change__03", "video_id": "uvo__0NUTcXmBjRs", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object A (boy in blue shirt riding a green toy tractor) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/be15fd0ea384e9856284/F0.jpg", "media/be15fd0ea384e9856284/F1.jpg", "media/be15fd0ea384e9856284/F2.jpg", "media/be15fd0ea384e9856284/F3.jpg", "media/be15fd0ea384e9856284/F4.jpg", "media/be15fd0ea384e9856284/F5.jpg", "media/be15fd0ea384e9856284/F6.jpg", "media/be15fd0ea384e9856284/F7.jpg", "media/be15fd0ea384e9856284/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (boy in blue shirt riding a green toy tractor) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a05cbd8f613c0dfd1b490cb16ec255edbb2d3fb3b8f6c49c60ba3a96084f077"}
{"qa_id": "uvo__0NUTcXmBjRs__object_world_motion_direction__04", "video_id": "uvo__0NUTcXmBjRs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object B (green ride-on toy tractor with yellow wheel rims) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "left", "right", "forward", "up"], "frame_paths": ["media/dd930737d15c2dfbf136/F0.jpg", "media/dd930737d15c2dfbf136/F1.jpg", "media/dd930737d15c2dfbf136/F2.jpg", "media/dd930737d15c2dfbf136/F3.jpg", "media/dd930737d15c2dfbf136/F4.jpg", "media/dd930737d15c2dfbf136/F5.jpg", "media/dd930737d15c2dfbf136/F6.jpg", "media/dd930737d15c2dfbf136/F7.jpg", "media/dd930737d15c2dfbf136/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object B (green ride-on toy tractor with yellow wheel rims) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. left\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "77c1b79257ba3146798ae07b3b8179fefba269a247572ef897f71543a1e6cee4"}
{"qa_id": "uvo__0NUTcXmBjRs__camera_object_distance_change__05", "video_id": "uvo__0NUTcXmBjRs", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, did object B (green ride-on toy tractor with yellow wheel rims) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/dd930737d15c2dfbf136/F0.jpg", "media/dd930737d15c2dfbf136/F1.jpg", "media/dd930737d15c2dfbf136/F2.jpg", "media/dd930737d15c2dfbf136/F3.jpg", "media/dd930737d15c2dfbf136/F4.jpg", "media/dd930737d15c2dfbf136/F5.jpg", "media/dd930737d15c2dfbf136/F6.jpg", "media/dd930737d15c2dfbf136/F7.jpg", "media/dd930737d15c2dfbf136/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object B (green ride-on toy tractor with yellow wheel rims) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4ad324db86bea1951abc3c2aaf41d33f1e39384e16117cf2b7132970adcbf894"}
{"qa_id": "uvo__0OFE-Ics0YU__camera_view_direction_change__00", "video_id": "uvo__0OFE-Ics0YU", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/ea479f60e641e38b9d8e/F0.jpg", "media/ea479f60e641e38b9d8e/F1.jpg", "media/ea479f60e641e38b9d8e/F2.jpg", "media/ea479f60e641e38b9d8e/F3.jpg", "media/ea479f60e641e38b9d8e/F4.jpg", "media/ea479f60e641e38b9d8e/F5.jpg", "media/ea479f60e641e38b9d8e/F6.jpg", "media/ea479f60e641e38b9d8e/F7.jpg", "media/ea479f60e641e38b9d8e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "133d8ba09663219319728ae6a225d6ed64a3b3eba34259b3eef49bdce9da79e9"}
{"qa_id": "uvo__0OVuyIAduMI__camera_view_direction_change__00", "video_id": "uvo__0OVuyIAduMI", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/7bf68da07c3a69004d2e/F0.jpg", "media/7bf68da07c3a69004d2e/F1.jpg", "media/7bf68da07c3a69004d2e/F2.jpg", "media/7bf68da07c3a69004d2e/F3.jpg", "media/7bf68da07c3a69004d2e/F4.jpg", "media/7bf68da07c3a69004d2e/F5.jpg", "media/7bf68da07c3a69004d2e/F6.jpg", "media/7bf68da07c3a69004d2e/F7.jpg", "media/7bf68da07c3a69004d2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "uvo__0P30A-kxQGo__camera_view_direction_change__00", "video_id": "uvo__0P30A-kxQGo", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/9308abc06dd0af587f81/F0.jpg", "media/9308abc06dd0af587f81/F1.jpg", "media/9308abc06dd0af587f81/F2.jpg", "media/9308abc06dd0af587f81/F3.jpg", "media/9308abc06dd0af587f81/F4.jpg", "media/9308abc06dd0af587f81/F5.jpg", "media/9308abc06dd0af587f81/F6.jpg", "media/9308abc06dd0af587f81/F7.jpg", "media/9308abc06dd0af587f81/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af75473fbfd63923e7a7dd9d31a1213d36f19af8df543626bae3d6af8373e9d0"}
{"qa_id": "uvo__0PTCyNDpb7c__camera_translation_direction__00", "video_id": "uvo__0PTCyNDpb7c", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/a1a6a66ab92074e11bff/F0.jpg", "media/a1a6a66ab92074e11bff/F1.jpg", "media/a1a6a66ab92074e11bff/F2.jpg", "media/a1a6a66ab92074e11bff/F3.jpg", "media/a1a6a66ab92074e11bff/F4.jpg", "media/a1a6a66ab92074e11bff/F5.jpg", "media/a1a6a66ab92074e11bff/F6.jpg", "media/a1a6a66ab92074e11bff/F7.jpg", "media/a1a6a66ab92074e11bff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0886a6abac6bc66024778fe6aff54c577c7b6df7bc12b2af89f7e24a94ba3b8"}
{"qa_id": "uvo__0PX-e1Vszs4__camera_view_direction_change__00", "video_id": "uvo__0PX-e1Vszs4", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/c857e035ecbc02d68ab5/F0.jpg", "media/c857e035ecbc02d68ab5/F1.jpg", "media/c857e035ecbc02d68ab5/F2.jpg", "media/c857e035ecbc02d68ab5/F3.jpg", "media/c857e035ecbc02d68ab5/F4.jpg", "media/c857e035ecbc02d68ab5/F5.jpg", "media/c857e035ecbc02d68ab5/F6.jpg", "media/c857e035ecbc02d68ab5/F7.jpg", "media/c857e035ecbc02d68ab5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5b8eb40004133ff943911c0e3dfad000929ebf881c834672ee889935c24142f"}
{"qa_id": "uvo__0Q5bI7XJYuE__object_world_motion_direction__00", "video_id": "uvo__0Q5bI7XJYuE", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person standing on the track) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "right", "left", "up"], "frame_paths": ["media/9e40ff1a1bafb721515e/F0.jpg", "media/9e40ff1a1bafb721515e/F1.jpg", "media/9e40ff1a1bafb721515e/F2.jpg", "media/9e40ff1a1bafb721515e/F3.jpg", "media/9e40ff1a1bafb721515e/F4.jpg", "media/9e40ff1a1bafb721515e/F5.jpg", "media/9e40ff1a1bafb721515e/F6.jpg", "media/9e40ff1a1bafb721515e/F7.jpg", "media/9e40ff1a1bafb721515e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person standing on the track) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "81b621b0d8619b4b01e344d4e82d63790d4bf59e5f5c19ff13bc3461cd8fb62e"}
{"qa_id": "uvo__0QUYujr9t54__camera_translation_direction__00", "video_id": "uvo__0QUYujr9t54", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "backward", "up"], "frame_paths": ["media/dfdbe0c6663b9fbccfbe/F0.jpg", "media/dfdbe0c6663b9fbccfbe/F1.jpg", "media/dfdbe0c6663b9fbccfbe/F2.jpg", "media/dfdbe0c6663b9fbccfbe/F3.jpg", "media/dfdbe0c6663b9fbccfbe/F4.jpg", "media/dfdbe0c6663b9fbccfbe/F5.jpg", "media/dfdbe0c6663b9fbccfbe/F6.jpg", "media/dfdbe0c6663b9fbccfbe/F7.jpg", "media/dfdbe0c6663b9fbccfbe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c07bb633b5e6e39a26f5a6c72ce899eca05b36c67248e8c83de3f1305807b02f"}
{"qa_id": "uvo__0R47I9RNtDY__camera_translation_direction__00", "video_id": "uvo__0R47I9RNtDY", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "right", "down", "backward", "forward"], "frame_paths": ["media/1f30bca8c6c3d2ff7112/F0.jpg", "media/1f30bca8c6c3d2ff7112/F1.jpg", "media/1f30bca8c6c3d2ff7112/F2.jpg", "media/1f30bca8c6c3d2ff7112/F3.jpg", "media/1f30bca8c6c3d2ff7112/F4.jpg", "media/1f30bca8c6c3d2ff7112/F5.jpg", "media/1f30bca8c6c3d2ff7112/F6.jpg", "media/1f30bca8c6c3d2ff7112/F7.jpg", "media/1f30bca8c6c3d2ff7112/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. right\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48b9cc420414da497eaa020494883029bb4028753b383f0cf00fa6b5b3ceabc0"}
{"qa_id": "uvo__0R47I9RNtDY__camera_view_direction_change__01", "video_id": "uvo__0R47I9RNtDY", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/1f30bca8c6c3d2ff7112/F0.jpg", "media/1f30bca8c6c3d2ff7112/F1.jpg", "media/1f30bca8c6c3d2ff7112/F2.jpg", "media/1f30bca8c6c3d2ff7112/F3.jpg", "media/1f30bca8c6c3d2ff7112/F4.jpg", "media/1f30bca8c6c3d2ff7112/F5.jpg", "media/1f30bca8c6c3d2ff7112/F6.jpg", "media/1f30bca8c6c3d2ff7112/F7.jpg", "media/1f30bca8c6c3d2ff7112/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f14fc5e0ad0c8f8b9b14b581d9cf65afb53e6a809c32e6f55a15fde4381e646"}
{"qa_id": "uvo__0Rr9z282Atw__object_world_motion_direction__00", "video_id": "uvo__0Rr9z282Atw", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in black jacket sitting on the snow) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/5814b0b4cf4b3039be56/F0.jpg", "media/5814b0b4cf4b3039be56/F1.jpg", "media/5814b0b4cf4b3039be56/F2.jpg", "media/5814b0b4cf4b3039be56/F3.jpg", "media/5814b0b4cf4b3039be56/F4.jpg", "media/5814b0b4cf4b3039be56/F5.jpg", "media/5814b0b4cf4b3039be56/F6.jpg", "media/5814b0b4cf4b3039be56/F7.jpg", "media/5814b0b4cf4b3039be56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in black jacket sitting on the snow) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "07fa1970529d0b7fa43ca320e460afc87397b887cbcc47cb470cc368a322833a"}
{"qa_id": "uvo__0Rr9z282Atw__camera_object_distance_change__01", "video_id": "uvo__0Rr9z282Atw", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, did object A (person in black jacket sitting on the snow) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5814b0b4cf4b3039be56/F0.jpg", "media/5814b0b4cf4b3039be56/F1.jpg", "media/5814b0b4cf4b3039be56/F2.jpg", "media/5814b0b4cf4b3039be56/F3.jpg", "media/5814b0b4cf4b3039be56/F4.jpg", "media/5814b0b4cf4b3039be56/F5.jpg", "media/5814b0b4cf4b3039be56/F6.jpg", "media/5814b0b4cf4b3039be56/F7.jpg", "media/5814b0b4cf4b3039be56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (person in black jacket sitting on the snow) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "946c3efc5d360cc9ceb29109669051c9bdfe165c54a8279e2cd6463c8160fefd"}
{"qa_id": "uvo__0SHRH4pN8h8__camera_view_direction_change__00", "video_id": "uvo__0SHRH4pN8h8", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/03225e5a59f17ef202f2/F0.jpg", "media/03225e5a59f17ef202f2/F1.jpg", "media/03225e5a59f17ef202f2/F2.jpg", "media/03225e5a59f17ef202f2/F3.jpg", "media/03225e5a59f17ef202f2/F4.jpg", "media/03225e5a59f17ef202f2/F5.jpg", "media/03225e5a59f17ef202f2/F6.jpg", "media/03225e5a59f17ef202f2/F7.jpg", "media/03225e5a59f17ef202f2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "daea01c2afd950f5019daa56e0ce06fa0f286d03575f5bdae183e746898299dc"}
{"qa_id": "uvo__0SfGmvlluvs__camera_view_direction_change__00", "video_id": "uvo__0SfGmvlluvs", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/ece5af36517ab6801d9a/F0.jpg", "media/ece5af36517ab6801d9a/F1.jpg", "media/ece5af36517ab6801d9a/F2.jpg", "media/ece5af36517ab6801d9a/F3.jpg", "media/ece5af36517ab6801d9a/F4.jpg", "media/ece5af36517ab6801d9a/F5.jpg", "media/ece5af36517ab6801d9a/F6.jpg", "media/ece5af36517ab6801d9a/F7.jpg", "media/ece5af36517ab6801d9a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66d7737839a5f00e56636a9d543886e534296c607206cb52be197319a9f92f0b"}
{"qa_id": "uvo__0U5eS-mrCLo__camera_translation_direction__00", "video_id": "uvo__0U5eS-mrCLo", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "forward", "backward"], "frame_paths": ["media/ae91ffa1951c03a56fd7/F0.jpg", "media/ae91ffa1951c03a56fd7/F1.jpg", "media/ae91ffa1951c03a56fd7/F2.jpg", "media/ae91ffa1951c03a56fd7/F3.jpg", "media/ae91ffa1951c03a56fd7/F4.jpg", "media/ae91ffa1951c03a56fd7/F5.jpg", "media/ae91ffa1951c03a56fd7/F6.jpg", "media/ae91ffa1951c03a56fd7/F7.jpg", "media/ae91ffa1951c03a56fd7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ddf3ab829d45c2c73f96bb1ff07eb16f0ac635c4993b505ab4722bac34ac4967"}
{"qa_id": "uvo__0UgyqJtni6g__camera_view_direction_change__00", "video_id": "uvo__0UgyqJtni6g", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/a1111989faf8da2549d1/F0.jpg", "media/a1111989faf8da2549d1/F1.jpg", "media/a1111989faf8da2549d1/F2.jpg", "media/a1111989faf8da2549d1/F3.jpg", "media/a1111989faf8da2549d1/F4.jpg", "media/a1111989faf8da2549d1/F5.jpg", "media/a1111989faf8da2549d1/F6.jpg", "media/a1111989faf8da2549d1/F7.jpg", "media/a1111989faf8da2549d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "27a943f0f444c042e7d4091ef447ee9c8b608b83471e530bb9a933f2310b4a46"}
{"qa_id": "uvo__0VzddQUqVCQ__camera_view_direction_change__00", "video_id": "uvo__0VzddQUqVCQ", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/00d84bbdbdb290367ac7/F0.jpg", "media/00d84bbdbdb290367ac7/F1.jpg", "media/00d84bbdbdb290367ac7/F2.jpg", "media/00d84bbdbdb290367ac7/F3.jpg", "media/00d84bbdbdb290367ac7/F4.jpg", "media/00d84bbdbdb290367ac7/F5.jpg", "media/00d84bbdbdb290367ac7/F6.jpg", "media/00d84bbdbdb290367ac7/F7.jpg", "media/00d84bbdbdb290367ac7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a125cfc0f77d2ae9b2e6bc05b025870e8cceaf01ceee8bce113b53b93422203e"}
{"qa_id": "uvo__0WaDbpdRZGc__camera_translation_direction__00", "video_id": "uvo__0WaDbpdRZGc", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/53c561095510949ed87e/F0.jpg", "media/53c561095510949ed87e/F1.jpg", "media/53c561095510949ed87e/F2.jpg", "media/53c561095510949ed87e/F3.jpg", "media/53c561095510949ed87e/F4.jpg", "media/53c561095510949ed87e/F5.jpg", "media/53c561095510949ed87e/F6.jpg", "media/53c561095510949ed87e/F7.jpg", "media/53c561095510949ed87e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5ad0329af0e3ff466b15a47480512de349bd9147f7c90f8a8e76ffb065b551c1"}
{"qa_id": "uvo__0WaDbpdRZGc__object_world_motion_direction__01", "video_id": "uvo__0WaDbpdRZGc", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (woman in red coat and black boots) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/a5f53e34cdfe10ed7a20/F0.jpg", "media/a5f53e34cdfe10ed7a20/F1.jpg", "media/a5f53e34cdfe10ed7a20/F2.jpg", "media/a5f53e34cdfe10ed7a20/F3.jpg", "media/a5f53e34cdfe10ed7a20/F4.jpg", "media/a5f53e34cdfe10ed7a20/F5.jpg", "media/a5f53e34cdfe10ed7a20/F6.jpg", "media/a5f53e34cdfe10ed7a20/F7.jpg", "media/a5f53e34cdfe10ed7a20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (woman in red coat and black boots) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "47894a6a05d3236df451a87d50cf1b97fd1379bdee8130e365baae1032eb390d"}
{"qa_id": "uvo__0WaDbpdRZGc__camera_object_distance_change__02", "video_id": "uvo__0WaDbpdRZGc", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object A (woman in red coat and black boots) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a5f53e34cdfe10ed7a20/F0.jpg", "media/a5f53e34cdfe10ed7a20/F1.jpg", "media/a5f53e34cdfe10ed7a20/F2.jpg", "media/a5f53e34cdfe10ed7a20/F3.jpg", "media/a5f53e34cdfe10ed7a20/F4.jpg", "media/a5f53e34cdfe10ed7a20/F5.jpg", "media/a5f53e34cdfe10ed7a20/F6.jpg", "media/a5f53e34cdfe10ed7a20/F7.jpg", "media/a5f53e34cdfe10ed7a20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (woman in red coat and black boots) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c38ac424de7633d2587191524d9e9024ad3ad4eee461bc2307b01c80af2739f8"}
{"qa_id": "uvo__0WaDbpdRZGc__object_world_motion_direction__03", "video_id": "uvo__0WaDbpdRZGc", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (child in pink plaid jacket and white skates) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/4ad827b344bbb3f8c8cf/F0.jpg", "media/4ad827b344bbb3f8c8cf/F1.jpg", "media/4ad827b344bbb3f8c8cf/F2.jpg", "media/4ad827b344bbb3f8c8cf/F3.jpg", "media/4ad827b344bbb3f8c8cf/F4.jpg", "media/4ad827b344bbb3f8c8cf/F5.jpg", "media/4ad827b344bbb3f8c8cf/F6.jpg", "media/4ad827b344bbb3f8c8cf/F7.jpg", "media/4ad827b344bbb3f8c8cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (child in pink plaid jacket and white skates) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e10c2f59c4efc9ff80441c07ee80a8e7e9adf43491cf31d52d4f5dae8c101012"}
{"qa_id": "uvo__0WaDbpdRZGc__camera_object_distance_change__04", "video_id": "uvo__0WaDbpdRZGc", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object B (child in pink plaid jacket and white skates) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/4ad827b344bbb3f8c8cf/F0.jpg", "media/4ad827b344bbb3f8c8cf/F1.jpg", "media/4ad827b344bbb3f8c8cf/F2.jpg", "media/4ad827b344bbb3f8c8cf/F3.jpg", "media/4ad827b344bbb3f8c8cf/F4.jpg", "media/4ad827b344bbb3f8c8cf/F5.jpg", "media/4ad827b344bbb3f8c8cf/F6.jpg", "media/4ad827b344bbb3f8c8cf/F7.jpg", "media/4ad827b344bbb3f8c8cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (child in pink plaid jacket and white skates) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3fc2d5a49177b6426dacc7166d4f701824223f4201133008257005aa045c852"}
{"qa_id": "uvo__0Y4AdXSciOs__camera_view_direction_change__00", "video_id": "uvo__0Y4AdXSciOs", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/ded9c8677d08687cfdc5/F0.jpg", "media/ded9c8677d08687cfdc5/F1.jpg", "media/ded9c8677d08687cfdc5/F2.jpg", "media/ded9c8677d08687cfdc5/F3.jpg", "media/ded9c8677d08687cfdc5/F4.jpg", "media/ded9c8677d08687cfdc5/F5.jpg", "media/ded9c8677d08687cfdc5/F6.jpg", "media/ded9c8677d08687cfdc5/F7.jpg", "media/ded9c8677d08687cfdc5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "uvo__0Y4AdXSciOs__camera_object_distance_change__01", "video_id": "uvo__0Y4AdXSciOs", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did object A (person wading in a river) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/64f1edb4d1dd1e2291c0/F0.jpg", "media/64f1edb4d1dd1e2291c0/F1.jpg", "media/64f1edb4d1dd1e2291c0/F2.jpg", "media/64f1edb4d1dd1e2291c0/F3.jpg", "media/64f1edb4d1dd1e2291c0/F4.jpg", "media/64f1edb4d1dd1e2291c0/F5.jpg", "media/64f1edb4d1dd1e2291c0/F6.jpg", "media/64f1edb4d1dd1e2291c0/F7.jpg", "media/64f1edb4d1dd1e2291c0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (person wading in a river) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "237d288da3e684e6b118508509aef1b1f84fa306d6514e889abeb68f406b64f6"}
{"qa_id": "uvo__0_tduei7o5g__camera_object_distance_change__00", "video_id": "uvo__0_tduei7o5g", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did object A (young man with short dark hair) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/34cda2a6a5443735e69c/F0.jpg", "media/34cda2a6a5443735e69c/F1.jpg", "media/34cda2a6a5443735e69c/F2.jpg", "media/34cda2a6a5443735e69c/F3.jpg", "media/34cda2a6a5443735e69c/F4.jpg", "media/34cda2a6a5443735e69c/F5.jpg", "media/34cda2a6a5443735e69c/F6.jpg", "media/34cda2a6a5443735e69c/F7.jpg", "media/34cda2a6a5443735e69c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (young man with short dark hair) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5b9c83994c154912b7ee4a2f40d8851c9410088d673a1782cd8c27467958afe9"}
{"qa_id": "uvo__0aQFYMZ98Rc__object_world_motion_direction__00", "video_id": "uvo__0aQFYMZ98Rc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in black shirt and shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "up", "backward", "right", "down"], "frame_paths": ["media/b5eb04e79a6c8dcca7c9/F0.jpg", "media/b5eb04e79a6c8dcca7c9/F1.jpg", "media/b5eb04e79a6c8dcca7c9/F2.jpg", "media/b5eb04e79a6c8dcca7c9/F3.jpg", "media/b5eb04e79a6c8dcca7c9/F4.jpg", "media/b5eb04e79a6c8dcca7c9/F5.jpg", "media/b5eb04e79a6c8dcca7c9/F6.jpg", "media/b5eb04e79a6c8dcca7c9/F7.jpg", "media/b5eb04e79a6c8dcca7c9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in black shirt and shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. up\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5316b4a4448510444b5eb6ada2f82772726a25b214537b5bd3700adff7a856f4"}
{"qa_id": "uvo__0bjuCXoqRAk__camera_view_direction_change__00", "video_id": "uvo__0bjuCXoqRAk", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/9d8ff2357e2d7a59f585/F0.jpg", "media/9d8ff2357e2d7a59f585/F1.jpg", "media/9d8ff2357e2d7a59f585/F2.jpg", "media/9d8ff2357e2d7a59f585/F3.jpg", "media/9d8ff2357e2d7a59f585/F4.jpg", "media/9d8ff2357e2d7a59f585/F5.jpg", "media/9d8ff2357e2d7a59f585/F6.jpg", "media/9d8ff2357e2d7a59f585/F7.jpg", "media/9d8ff2357e2d7a59f585/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "133d8ba09663219319728ae6a225d6ed64a3b3eba34259b3eef49bdce9da79e9"}
{"qa_id": "uvo__0bkmUXjwKaA__camera_view_direction_change__00", "video_id": "uvo__0bkmUXjwKaA", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/9df21c9246b2afc56d4c/F0.jpg", "media/9df21c9246b2afc56d4c/F1.jpg", "media/9df21c9246b2afc56d4c/F2.jpg", "media/9df21c9246b2afc56d4c/F3.jpg", "media/9df21c9246b2afc56d4c/F4.jpg", "media/9df21c9246b2afc56d4c/F5.jpg", "media/9df21c9246b2afc56d4c/F6.jpg", "media/9df21c9246b2afc56d4c/F7.jpg", "media/9df21c9246b2afc56d4c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b753cba65d52068fe6729fa64acba3f9e6e5941e267c052876d88c95ba60a2f1"}
{"qa_id": "uvo__0bkmUXjwKaA__object_world_motion_direction__01", "video_id": "uvo__0bkmUXjwKaA", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (person in dark clothing standing on the ice) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "up", "right", "forward"], "frame_paths": ["media/f947eebd9214245513c4/F0.jpg", "media/f947eebd9214245513c4/F1.jpg", "media/f947eebd9214245513c4/F2.jpg", "media/f947eebd9214245513c4/F3.jpg", "media/f947eebd9214245513c4/F4.jpg", "media/f947eebd9214245513c4/F5.jpg", "media/f947eebd9214245513c4/F6.jpg", "media/f947eebd9214245513c4/F7.jpg", "media/f947eebd9214245513c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (person in dark clothing standing on the ice) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a77c333b87bce1131c0843d02794865b9c66aa12e7d4e0a0e993ddef96eec697"}
{"qa_id": "uvo__0bkmUXjwKaA__object_pair_distance_change__02", "video_id": "uvo__0bkmUXjwKaA", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F5, did the 3D distance between object A (dark bird in flight) and object B (person in dark clothing standing on the ice) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/8255cdacabc4d31f47a2/F0.jpg", "media/8255cdacabc4d31f47a2/F1.jpg", "media/8255cdacabc4d31f47a2/F2.jpg", "media/8255cdacabc4d31f47a2/F3.jpg", "media/8255cdacabc4d31f47a2/F4.jpg", "media/8255cdacabc4d31f47a2/F5.jpg", "media/8255cdacabc4d31f47a2/F6.jpg", "media/8255cdacabc4d31f47a2/F7.jpg", "media/8255cdacabc4d31f47a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (dark bird in flight) and object B (person in dark clothing standing on the ice) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "258e93fab8b54f9ccee0a4c9b901369a5d350ef4993b998f9dc4cff6e7643694"}
{"qa_id": "uvo__0bkmUXjwKaA__object_travel_distance_comparison__03", "video_id": "uvo__0bkmUXjwKaA", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark bird in flight) or B (person in dark clothing standing on the ice)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/8255cdacabc4d31f47a2/F0.jpg", "media/8255cdacabc4d31f47a2/F1.jpg", "media/8255cdacabc4d31f47a2/F2.jpg", "media/8255cdacabc4d31f47a2/F3.jpg", "media/8255cdacabc4d31f47a2/F4.jpg", "media/8255cdacabc4d31f47a2/F5.jpg", "media/8255cdacabc4d31f47a2/F6.jpg", "media/8255cdacabc4d31f47a2/F7.jpg", "media/8255cdacabc4d31f47a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (dark bird in flight) or B (person in dark clothing standing on the ice)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "17e76c2183589d895a8a38acf6413864088ce1e667cc9daa778e76a8b6d693af"}
{"qa_id": "uvo__0cGDbQtYXms__camera_view_direction_change__00", "video_id": "uvo__0cGDbQtYXms", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/122ce622519b122f41cb/F0.jpg", "media/122ce622519b122f41cb/F1.jpg", "media/122ce622519b122f41cb/F2.jpg", "media/122ce622519b122f41cb/F3.jpg", "media/122ce622519b122f41cb/F4.jpg", "media/122ce622519b122f41cb/F5.jpg", "media/122ce622519b122f41cb/F6.jpg", "media/122ce622519b122f41cb/F7.jpg", "media/122ce622519b122f41cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "uvo__0cR5ap1n5Uo__camera_view_direction_change__00", "video_id": "uvo__0cR5ap1n5Uo", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/5862223a819c64b4602d/F0.jpg", "media/5862223a819c64b4602d/F1.jpg", "media/5862223a819c64b4602d/F2.jpg", "media/5862223a819c64b4602d/F3.jpg", "media/5862223a819c64b4602d/F4.jpg", "media/5862223a819c64b4602d/F5.jpg", "media/5862223a819c64b4602d/F6.jpg", "media/5862223a819c64b4602d/F7.jpg", "media/5862223a819c64b4602d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "uvo__0cSmJPgrEXE__camera_view_direction_change__00", "video_id": "uvo__0cSmJPgrEXE", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/bfb881c52a88e6a3f928/F0.jpg", "media/bfb881c52a88e6a3f928/F1.jpg", "media/bfb881c52a88e6a3f928/F2.jpg", "media/bfb881c52a88e6a3f928/F3.jpg", "media/bfb881c52a88e6a3f928/F4.jpg", "media/bfb881c52a88e6a3f928/F5.jpg", "media/bfb881c52a88e6a3f928/F6.jpg", "media/bfb881c52a88e6a3f928/F7.jpg", "media/bfb881c52a88e6a3f928/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8072d60bbd8d6d0e92c754b79a41d83ad026a2f6e6d929da31479a7692b87ed"}
{"qa_id": "uvo__0cSmJPgrEXE__object_world_motion_direction__01", "video_id": "uvo__0cSmJPgrEXE", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (boy in grey shirt and black pants with green stripe) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/67d6cd134f41e1ddd4ec/F0.jpg", "media/67d6cd134f41e1ddd4ec/F1.jpg", "media/67d6cd134f41e1ddd4ec/F2.jpg", "media/67d6cd134f41e1ddd4ec/F3.jpg", "media/67d6cd134f41e1ddd4ec/F4.jpg", "media/67d6cd134f41e1ddd4ec/F5.jpg", "media/67d6cd134f41e1ddd4ec/F6.jpg", "media/67d6cd134f41e1ddd4ec/F7.jpg", "media/67d6cd134f41e1ddd4ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (boy in grey shirt and black pants with green stripe) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2924da090a6da679119d09c62c967042b1a2552a0c365ed58ee43fb07ffaf6b3"}
{"qa_id": "uvo__0cSmJPgrEXE__camera_object_distance_change__02", "video_id": "uvo__0cSmJPgrEXE", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, did object A (boy in grey shirt and black pants with green stripe) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/67d6cd134f41e1ddd4ec/F0.jpg", "media/67d6cd134f41e1ddd4ec/F1.jpg", "media/67d6cd134f41e1ddd4ec/F2.jpg", "media/67d6cd134f41e1ddd4ec/F3.jpg", "media/67d6cd134f41e1ddd4ec/F4.jpg", "media/67d6cd134f41e1ddd4ec/F5.jpg", "media/67d6cd134f41e1ddd4ec/F6.jpg", "media/67d6cd134f41e1ddd4ec/F7.jpg", "media/67d6cd134f41e1ddd4ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (boy in grey shirt and black pants with green stripe) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b9d8d17a0ee21afe2f153d0775a2634e8a961eae9b77945f3f7f8d0e4f76f107"}
{"qa_id": "uvo__0d5Y3zRJJkw__camera_view_direction_change__00", "video_id": "uvo__0d5Y3zRJJkw", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/85d6201cad5d37133346/F0.jpg", "media/85d6201cad5d37133346/F1.jpg", "media/85d6201cad5d37133346/F2.jpg", "media/85d6201cad5d37133346/F3.jpg", "media/85d6201cad5d37133346/F4.jpg", "media/85d6201cad5d37133346/F5.jpg", "media/85d6201cad5d37133346/F6.jpg", "media/85d6201cad5d37133346/F7.jpg", "media/85d6201cad5d37133346/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654016ae105affb53bc2face594c67e96f587f4cdc806a6f418076b30032d255"}
{"qa_id": "uvo__0d5Y3zRJJkw__object_world_motion_direction__01", "video_id": "uvo__0d5Y3zRJJkw", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (person riding a horse, wearing dark clothing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "backward", "forward", "up"], "frame_paths": ["media/60143604ecd1661f4513/F0.jpg", "media/60143604ecd1661f4513/F1.jpg", "media/60143604ecd1661f4513/F2.jpg", "media/60143604ecd1661f4513/F3.jpg", "media/60143604ecd1661f4513/F4.jpg", "media/60143604ecd1661f4513/F5.jpg", "media/60143604ecd1661f4513/F6.jpg", "media/60143604ecd1661f4513/F7.jpg", "media/60143604ecd1661f4513/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (person riding a horse, wearing dark clothing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "feb6d2b0a264ea8dfa4658d9381c2324c84c72af7d17989ba4d55e286ed04149"}
{"qa_id": "uvo__0d5Y3zRJJkw__object_world_motion_direction__02", "video_id": "uvo__0d5Y3zRJJkw", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (brown horse with a rider) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "up", "forward", "backward"], "frame_paths": ["media/834f3e20ea95b9214dee/F0.jpg", "media/834f3e20ea95b9214dee/F1.jpg", "media/834f3e20ea95b9214dee/F2.jpg", "media/834f3e20ea95b9214dee/F3.jpg", "media/834f3e20ea95b9214dee/F4.jpg", "media/834f3e20ea95b9214dee/F5.jpg", "media/834f3e20ea95b9214dee/F6.jpg", "media/834f3e20ea95b9214dee/F7.jpg", "media/834f3e20ea95b9214dee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (brown horse with a rider) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40b4b904a81faeb98b98d068d03c497eb9662c670d7fef8a405161772282b886"}
{"qa_id": "uvo__0dDzds7o0BM__camera_translation_direction__00", "video_id": "uvo__0dDzds7o0BM", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/b3d4a420e3e199fdfc55/F0.jpg", "media/b3d4a420e3e199fdfc55/F1.jpg", "media/b3d4a420e3e199fdfc55/F2.jpg", "media/b3d4a420e3e199fdfc55/F3.jpg", "media/b3d4a420e3e199fdfc55/F4.jpg", "media/b3d4a420e3e199fdfc55/F5.jpg", "media/b3d4a420e3e199fdfc55/F6.jpg", "media/b3d4a420e3e199fdfc55/F7.jpg", "media/b3d4a420e3e199fdfc55/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7cd6242de61f808b225bbfb91276cb246620db5d9ae1d9d7dbba1de41a7fa57d"}
{"qa_id": "uvo__0dDzds7o0BM__camera_object_distance_change__01", "video_id": "uvo__0dDzds7o0BM", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, did object A (person skiing on a snowy slope) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ed2ffe12697e011697ea/F0.jpg", "media/ed2ffe12697e011697ea/F1.jpg", "media/ed2ffe12697e011697ea/F2.jpg", "media/ed2ffe12697e011697ea/F3.jpg", "media/ed2ffe12697e011697ea/F4.jpg", "media/ed2ffe12697e011697ea/F5.jpg", "media/ed2ffe12697e011697ea/F6.jpg", "media/ed2ffe12697e011697ea/F7.jpg", "media/ed2ffe12697e011697ea/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (person skiing on a snowy slope) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "59782afb39e34774c9be80d4b66d0835b6ac9c9bd52e1112ff69a5f7451f7a6e"}
{"qa_id": "uvo__0e2OdCAOwe8__object_world_motion_direction__00", "video_id": "uvo__0e2OdCAOwe8", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (silhouette of a horse with a saddle) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/65b852294367d8df223f/F0.jpg", "media/65b852294367d8df223f/F1.jpg", "media/65b852294367d8df223f/F2.jpg", "media/65b852294367d8df223f/F3.jpg", "media/65b852294367d8df223f/F4.jpg", "media/65b852294367d8df223f/F5.jpg", "media/65b852294367d8df223f/F6.jpg", "media/65b852294367d8df223f/F7.jpg", "media/65b852294367d8df223f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (silhouette of a horse with a saddle) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a4bee2092e7d4d89fa0ca453ec7ccbf123f1d5c9ebcfb9ab27e6763c64eac8ed"}
{"qa_id": "uvo__0eRkpTGq5pA__object_world_motion_direction__00", "video_id": "uvo__0eRkpTGq5pA", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (man in white t-shirt and black shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "backward", "forward"], "frame_paths": ["media/8f6d8d4fc3e4bb136129/F0.jpg", "media/8f6d8d4fc3e4bb136129/F1.jpg", "media/8f6d8d4fc3e4bb136129/F2.jpg", "media/8f6d8d4fc3e4bb136129/F3.jpg", "media/8f6d8d4fc3e4bb136129/F4.jpg", "media/8f6d8d4fc3e4bb136129/F5.jpg", "media/8f6d8d4fc3e4bb136129/F6.jpg", "media/8f6d8d4fc3e4bb136129/F7.jpg", "media/8f6d8d4fc3e4bb136129/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (man in white t-shirt and black shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4def67affd05cc884ad2fc87a683789bcb1bbf2760c0b8c6f1998dcdbfa91e49"}
{"qa_id": "uvo__0eRkpTGq5pA__camera_object_distance_change__01", "video_id": "uvo__0eRkpTGq5pA", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (man in white t-shirt and black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/8f6d8d4fc3e4bb136129/F0.jpg", "media/8f6d8d4fc3e4bb136129/F1.jpg", "media/8f6d8d4fc3e4bb136129/F2.jpg", "media/8f6d8d4fc3e4bb136129/F3.jpg", "media/8f6d8d4fc3e4bb136129/F4.jpg", "media/8f6d8d4fc3e4bb136129/F5.jpg", "media/8f6d8d4fc3e4bb136129/F6.jpg", "media/8f6d8d4fc3e4bb136129/F7.jpg", "media/8f6d8d4fc3e4bb136129/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (man in white t-shirt and black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "20e8dbbd8617612b9077397e8607c614011a159647fae3b9c68996d16cf1ef8e"}
{"qa_id": "uvo__0eRkpTGq5pA__camera_object_distance_change__02", "video_id": "uvo__0eRkpTGq5pA", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object B (man in dark blue long-sleeve shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/952252dc57b2b5c0b8c0/F0.jpg", "media/952252dc57b2b5c0b8c0/F1.jpg", "media/952252dc57b2b5c0b8c0/F2.jpg", "media/952252dc57b2b5c0b8c0/F3.jpg", "media/952252dc57b2b5c0b8c0/F4.jpg", "media/952252dc57b2b5c0b8c0/F5.jpg", "media/952252dc57b2b5c0b8c0/F6.jpg", "media/952252dc57b2b5c0b8c0/F7.jpg", "media/952252dc57b2b5c0b8c0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (man in dark blue long-sleeve shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc548a9455c6007e22e4b205fc406d9906cee28a3ec1eb1462b85f37cdc7c77c"}
{"qa_id": "uvo__0eSFmX-NTIU__object_world_motion_direction__00", "video_id": "uvo__0eSFmX-NTIU", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (large dark elephant with riders on its back) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "left", "forward", "right", "down", "backward"], "frame_paths": ["media/21fbb5194562500c7b8c/F0.jpg", "media/21fbb5194562500c7b8c/F1.jpg", "media/21fbb5194562500c7b8c/F2.jpg", "media/21fbb5194562500c7b8c/F3.jpg", "media/21fbb5194562500c7b8c/F4.jpg", "media/21fbb5194562500c7b8c/F5.jpg", "media/21fbb5194562500c7b8c/F6.jpg", "media/21fbb5194562500c7b8c/F7.jpg", "media/21fbb5194562500c7b8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (large dark elephant with riders on its back) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. forward\nD. right\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b045bcd00d4f143bf443a912e8bf6a240c55e9e183da0cd0bb7154da651de342"}
{"qa_id": "uvo__0eSFmX-NTIU__camera_object_distance_change__01", "video_id": "uvo__0eSFmX-NTIU", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (large dark elephant with riders on its back) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/21fbb5194562500c7b8c/F0.jpg", "media/21fbb5194562500c7b8c/F1.jpg", "media/21fbb5194562500c7b8c/F2.jpg", "media/21fbb5194562500c7b8c/F3.jpg", "media/21fbb5194562500c7b8c/F4.jpg", "media/21fbb5194562500c7b8c/F5.jpg", "media/21fbb5194562500c7b8c/F6.jpg", "media/21fbb5194562500c7b8c/F7.jpg", "media/21fbb5194562500c7b8c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (large dark elephant with riders on its back) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ad5be9391a89d566a0c06807626e37ab8d13e662ce2b620c7c93100801f97e61"}
{"qa_id": "uvo__0eSFmX-NTIU__object_pair_distance_change__02", "video_id": "uvo__0eSFmX-NTIU", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did the 3D distance between object A (large dark elephant with riders on its back) and object B (person in light blue shirt standing near the elephant) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/a010047cf9ddc6f99035/F0.jpg", "media/a010047cf9ddc6f99035/F1.jpg", "media/a010047cf9ddc6f99035/F2.jpg", "media/a010047cf9ddc6f99035/F3.jpg", "media/a010047cf9ddc6f99035/F4.jpg", "media/a010047cf9ddc6f99035/F5.jpg", "media/a010047cf9ddc6f99035/F6.jpg", "media/a010047cf9ddc6f99035/F7.jpg", "media/a010047cf9ddc6f99035/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (large dark elephant with riders on its back) and object B (person in light blue shirt standing near the elephant) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "81a2e346c207f79c606d08ae31d4b8a73c320a33d221146d5ad2f3f11c3d8d03"}
{"qa_id": "uvo__0e_dwOfB_bU__camera_view_direction_change__00", "video_id": "uvo__0e_dwOfB_bU", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/a3b3894b20f4886364c7/F0.jpg", "media/a3b3894b20f4886364c7/F1.jpg", "media/a3b3894b20f4886364c7/F2.jpg", "media/a3b3894b20f4886364c7/F3.jpg", "media/a3b3894b20f4886364c7/F4.jpg", "media/a3b3894b20f4886364c7/F5.jpg", "media/a3b3894b20f4886364c7/F6.jpg", "media/a3b3894b20f4886364c7/F7.jpg", "media/a3b3894b20f4886364c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aae452319b2f4945644936359ec7d3add997dd5675fbc7ea01e8e75c706f84dd"}
{"qa_id": "uvo__0ewQXPz9kWk__camera_translation_direction__00", "video_id": "uvo__0ewQXPz9kWk", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "down", "backward", "up", "right"], "frame_paths": ["media/15f20f540101dc90ebdc/F0.jpg", "media/15f20f540101dc90ebdc/F1.jpg", "media/15f20f540101dc90ebdc/F2.jpg", "media/15f20f540101dc90ebdc/F3.jpg", "media/15f20f540101dc90ebdc/F4.jpg", "media/15f20f540101dc90ebdc/F5.jpg", "media/15f20f540101dc90ebdc/F6.jpg", "media/15f20f540101dc90ebdc/F7.jpg", "media/15f20f540101dc90ebdc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. down\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9608a1283bd52644e960193b54c3c0e91aff9ad2da26c4e8266e1b4ccc1d4153"}
{"qa_id": "uvo__0fItLr9CtVg__camera_view_direction_change__00", "video_id": "uvo__0fItLr9CtVg", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/5452175027b4309da98b/F0.jpg", "media/5452175027b4309da98b/F1.jpg", "media/5452175027b4309da98b/F2.jpg", "media/5452175027b4309da98b/F3.jpg", "media/5452175027b4309da98b/F4.jpg", "media/5452175027b4309da98b/F5.jpg", "media/5452175027b4309da98b/F6.jpg", "media/5452175027b4309da98b/F7.jpg", "media/5452175027b4309da98b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55c282928f8940355d2075dba86d8619c91a164d28fb231d532d13fe514a5de1"}
{"qa_id": "uvo__0hxF743aOl8__camera_view_direction_change__00", "video_id": "uvo__0hxF743aOl8", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/fa328e36898f213cfe7f/F0.jpg", "media/fa328e36898f213cfe7f/F1.jpg", "media/fa328e36898f213cfe7f/F2.jpg", "media/fa328e36898f213cfe7f/F3.jpg", "media/fa328e36898f213cfe7f/F4.jpg", "media/fa328e36898f213cfe7f/F5.jpg", "media/fa328e36898f213cfe7f/F6.jpg", "media/fa328e36898f213cfe7f/F7.jpg", "media/fa328e36898f213cfe7f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__0igSLIh3ntQ__camera_translation_direction__00", "video_id": "uvo__0igSLIh3ntQ", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "left", "forward", "backward"], "frame_paths": ["media/cb430241604f7a33664e/F0.jpg", "media/cb430241604f7a33664e/F1.jpg", "media/cb430241604f7a33664e/F2.jpg", "media/cb430241604f7a33664e/F3.jpg", "media/cb430241604f7a33664e/F4.jpg", "media/cb430241604f7a33664e/F5.jpg", "media/cb430241604f7a33664e/F6.jpg", "media/cb430241604f7a33664e/F7.jpg", "media/cb430241604f7a33664e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "12e59c80877a78e6ca45e9139ae89c79aaf7336131daca5714fbdc4873e2fa07"}
{"qa_id": "uvo__0igSLIh3ntQ__camera_view_direction_change__01", "video_id": "uvo__0igSLIh3ntQ", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/cb430241604f7a33664e/F0.jpg", "media/cb430241604f7a33664e/F1.jpg", "media/cb430241604f7a33664e/F2.jpg", "media/cb430241604f7a33664e/F3.jpg", "media/cb430241604f7a33664e/F4.jpg", "media/cb430241604f7a33664e/F5.jpg", "media/cb430241604f7a33664e/F6.jpg", "media/cb430241604f7a33664e/F7.jpg", "media/cb430241604f7a33664e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d93f29ad1c79e991b10ef9b3e54180f27685079141becefe88efd4a6946446eb"}
{"qa_id": "uvo__0igSLIh3ntQ__object_world_motion_direction__02", "video_id": "uvo__0igSLIh3ntQ", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (white horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/04dbb9dbcc32a8fc2b77/F0.jpg", "media/04dbb9dbcc32a8fc2b77/F1.jpg", "media/04dbb9dbcc32a8fc2b77/F2.jpg", "media/04dbb9dbcc32a8fc2b77/F3.jpg", "media/04dbb9dbcc32a8fc2b77/F4.jpg", "media/04dbb9dbcc32a8fc2b77/F5.jpg", "media/04dbb9dbcc32a8fc2b77/F6.jpg", "media/04dbb9dbcc32a8fc2b77/F7.jpg", "media/04dbb9dbcc32a8fc2b77/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (white horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fccf41034a5cbf494e53a87d70ef7e57c12943d2c6330e6ff19524a23deb9d5a"}
{"qa_id": "uvo__0igSLIh3ntQ__object_world_motion_direction__03", "video_id": "uvo__0igSLIh3ntQ", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object B (person in dark shirt riding the white horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/affe682ab7e9bc9981c6/F0.jpg", "media/affe682ab7e9bc9981c6/F1.jpg", "media/affe682ab7e9bc9981c6/F2.jpg", "media/affe682ab7e9bc9981c6/F3.jpg", "media/affe682ab7e9bc9981c6/F4.jpg", "media/affe682ab7e9bc9981c6/F5.jpg", "media/affe682ab7e9bc9981c6/F6.jpg", "media/affe682ab7e9bc9981c6/F7.jpg", "media/affe682ab7e9bc9981c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object B (person in dark shirt riding the white horse) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "90fa688e515c4728f87707da22203b68e9c8f2208e5d4450d34ee7f11c751fdf"}
{"qa_id": "uvo__0jFBHhjvbYs__camera_view_direction_change__00", "video_id": "uvo__0jFBHhjvbYs", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/1c14e727f7bc1f97ba40/F0.jpg", "media/1c14e727f7bc1f97ba40/F1.jpg", "media/1c14e727f7bc1f97ba40/F2.jpg", "media/1c14e727f7bc1f97ba40/F3.jpg", "media/1c14e727f7bc1f97ba40/F4.jpg", "media/1c14e727f7bc1f97ba40/F5.jpg", "media/1c14e727f7bc1f97ba40/F6.jpg", "media/1c14e727f7bc1f97ba40/F7.jpg", "media/1c14e727f7bc1f97ba40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4bb7ec21e095c5c46cce0347f69602ee73fee0909fed7189c01b64cbf085997"}
{"qa_id": "uvo__0jQaeta58yI__camera_translation_direction__00", "video_id": "uvo__0jQaeta58yI", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "left", "forward", "right"], "frame_paths": ["media/2434f3be18ae8a2db37c/F0.jpg", "media/2434f3be18ae8a2db37c/F1.jpg", "media/2434f3be18ae8a2db37c/F2.jpg", "media/2434f3be18ae8a2db37c/F3.jpg", "media/2434f3be18ae8a2db37c/F4.jpg", "media/2434f3be18ae8a2db37c/F5.jpg", "media/2434f3be18ae8a2db37c/F6.jpg", "media/2434f3be18ae8a2db37c/F7.jpg", "media/2434f3be18ae8a2db37c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. left\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "687fddb4b44a5e84a389e48bf4a915c2a8969de490fd487f61670284cdd5169f"}
{"qa_id": "uvo__0jQaeta58yI__object_world_motion_direction__01", "video_id": "uvo__0jQaeta58yI", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in white shirt riding a bicycle) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "left", "forward", "backward"], "frame_paths": ["media/51211a1f11887d36a821/F0.jpg", "media/51211a1f11887d36a821/F1.jpg", "media/51211a1f11887d36a821/F2.jpg", "media/51211a1f11887d36a821/F3.jpg", "media/51211a1f11887d36a821/F4.jpg", "media/51211a1f11887d36a821/F5.jpg", "media/51211a1f11887d36a821/F6.jpg", "media/51211a1f11887d36a821/F7.jpg", "media/51211a1f11887d36a821/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in white shirt riding a bicycle) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0590cf9efbea914963c1628b05fb0b9cfdad788e33a89deeec128baa73e33e8d"}
{"qa_id": "uvo__0jQaeta58yI__object_world_motion_direction__02", "video_id": "uvo__0jQaeta58yI", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (bicycle being ridden by a person in a white shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "forward", "down", "right"], "frame_paths": ["media/bc15b5ddaf243123ae1c/F0.jpg", "media/bc15b5ddaf243123ae1c/F1.jpg", "media/bc15b5ddaf243123ae1c/F2.jpg", "media/bc15b5ddaf243123ae1c/F3.jpg", "media/bc15b5ddaf243123ae1c/F4.jpg", "media/bc15b5ddaf243123ae1c/F5.jpg", "media/bc15b5ddaf243123ae1c/F6.jpg", "media/bc15b5ddaf243123ae1c/F7.jpg", "media/bc15b5ddaf243123ae1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (bicycle being ridden by a person in a white shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. forward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3dac9f9475fb9234d161a84c013fce20188249428df4b6085237e388c9625143"}
{"qa_id": "uvo__0kgm9iuzKew__camera_view_direction_change__00", "video_id": "uvo__0kgm9iuzKew", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/e258bcaa565d0ba3647f/F0.jpg", "media/e258bcaa565d0ba3647f/F1.jpg", "media/e258bcaa565d0ba3647f/F2.jpg", "media/e258bcaa565d0ba3647f/F3.jpg", "media/e258bcaa565d0ba3647f/F4.jpg", "media/e258bcaa565d0ba3647f/F5.jpg", "media/e258bcaa565d0ba3647f/F6.jpg", "media/e258bcaa565d0ba3647f/F7.jpg", "media/e258bcaa565d0ba3647f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__0kkPIwmcpOg__camera_translation_direction__00", "video_id": "uvo__0kkPIwmcpOg", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "down", "right", "up", "left"], "frame_paths": ["media/2b0c444a701bf46d6b33/F0.jpg", "media/2b0c444a701bf46d6b33/F1.jpg", "media/2b0c444a701bf46d6b33/F2.jpg", "media/2b0c444a701bf46d6b33/F3.jpg", "media/2b0c444a701bf46d6b33/F4.jpg", "media/2b0c444a701bf46d6b33/F5.jpg", "media/2b0c444a701bf46d6b33/F6.jpg", "media/2b0c444a701bf46d6b33/F7.jpg", "media/2b0c444a701bf46d6b33/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. down\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3f6d050a7067400b3ef92a6a4eb08e3ac6da4940b49586dcc322619976560792"}
{"qa_id": "uvo__0l8GZuA6QPo__camera_translation_direction__00", "video_id": "uvo__0l8GZuA6QPo", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/9c995d7df7ac954888dd/F0.jpg", "media/9c995d7df7ac954888dd/F1.jpg", "media/9c995d7df7ac954888dd/F2.jpg", "media/9c995d7df7ac954888dd/F3.jpg", "media/9c995d7df7ac954888dd/F4.jpg", "media/9c995d7df7ac954888dd/F5.jpg", "media/9c995d7df7ac954888dd/F6.jpg", "media/9c995d7df7ac954888dd/F7.jpg", "media/9c995d7df7ac954888dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "086b0590fcdf23510fe882bb8fbfe840657e96dbb235f3576ea9022182ca9d65"}
{"qa_id": "uvo__0lX9npPI08w__camera_view_direction_change__00", "video_id": "uvo__0lX9npPI08w", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/be360e7ec67fd3d1ca5f/F0.jpg", "media/be360e7ec67fd3d1ca5f/F1.jpg", "media/be360e7ec67fd3d1ca5f/F2.jpg", "media/be360e7ec67fd3d1ca5f/F3.jpg", "media/be360e7ec67fd3d1ca5f/F4.jpg", "media/be360e7ec67fd3d1ca5f/F5.jpg", "media/be360e7ec67fd3d1ca5f/F6.jpg", "media/be360e7ec67fd3d1ca5f/F7.jpg", "media/be360e7ec67fd3d1ca5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fec7177b28cd260bac8abbd97c394e95ca01afdba0cbfecdcc2d079d3b0a8ef6"}
{"qa_id": "uvo__0lp9mBNDAQM__camera_view_direction_change__00", "video_id": "uvo__0lp9mBNDAQM", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/bc727fecc65b8f83ecbb/F0.jpg", "media/bc727fecc65b8f83ecbb/F1.jpg", "media/bc727fecc65b8f83ecbb/F2.jpg", "media/bc727fecc65b8f83ecbb/F3.jpg", "media/bc727fecc65b8f83ecbb/F4.jpg", "media/bc727fecc65b8f83ecbb/F5.jpg", "media/bc727fecc65b8f83ecbb/F6.jpg", "media/bc727fecc65b8f83ecbb/F7.jpg", "media/bc727fecc65b8f83ecbb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "93626333f8ae00b8c5b7909c9540b956558a831446b9e15fc7ffbabc05439ca1"}
{"qa_id": "uvo__0nUL5Qi9k7Q__camera_view_direction_change__00", "video_id": "uvo__0nUL5Qi9k7Q", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/0ace218df0742e183f92/F0.jpg", "media/0ace218df0742e183f92/F1.jpg", "media/0ace218df0742e183f92/F2.jpg", "media/0ace218df0742e183f92/F3.jpg", "media/0ace218df0742e183f92/F4.jpg", "media/0ace218df0742e183f92/F5.jpg", "media/0ace218df0742e183f92/F6.jpg", "media/0ace218df0742e183f92/F7.jpg", "media/0ace218df0742e183f92/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "072f60231aa3802f6986b76870c84b4ad99855a926274c179ddcf45676cb1825"}
{"qa_id": "uvo__0nd-Gc3HkmU__camera_view_direction_change__00", "video_id": "uvo__0nd-Gc3HkmU", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/1ffe29a23bec2932e6ee/F0.jpg", "media/1ffe29a23bec2932e6ee/F1.jpg", "media/1ffe29a23bec2932e6ee/F2.jpg", "media/1ffe29a23bec2932e6ee/F3.jpg", "media/1ffe29a23bec2932e6ee/F4.jpg", "media/1ffe29a23bec2932e6ee/F5.jpg", "media/1ffe29a23bec2932e6ee/F6.jpg", "media/1ffe29a23bec2932e6ee/F7.jpg", "media/1ffe29a23bec2932e6ee/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "547ed7db89fa2dd82f0927c4be34faa90562781b9745f836becafedb1759aa3c"}
{"qa_id": "uvo__0oEt6N6POFI__camera_object_distance_change__00", "video_id": "uvo__0oEt6N6POFI", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (green tractor with a flatbed trailer) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/0db3dbe3b31e75fd98c6/F0.jpg", "media/0db3dbe3b31e75fd98c6/F1.jpg", "media/0db3dbe3b31e75fd98c6/F2.jpg", "media/0db3dbe3b31e75fd98c6/F3.jpg", "media/0db3dbe3b31e75fd98c6/F4.jpg", "media/0db3dbe3b31e75fd98c6/F5.jpg", "media/0db3dbe3b31e75fd98c6/F6.jpg", "media/0db3dbe3b31e75fd98c6/F7.jpg", "media/0db3dbe3b31e75fd98c6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (green tractor with a flatbed trailer) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0bfbc0a43544540dd3140df90cc11b3868792da8fabea571a13b8aecd70568e"}
{"qa_id": "uvo__0oMNrDyWUEo__camera_translation_direction__00", "video_id": "uvo__0oMNrDyWUEo", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "down", "up", "backward"], "frame_paths": ["media/e30199359ed09f1088b1/F0.jpg", "media/e30199359ed09f1088b1/F1.jpg", "media/e30199359ed09f1088b1/F2.jpg", "media/e30199359ed09f1088b1/F3.jpg", "media/e30199359ed09f1088b1/F4.jpg", "media/e30199359ed09f1088b1/F5.jpg", "media/e30199359ed09f1088b1/F6.jpg", "media/e30199359ed09f1088b1/F7.jpg", "media/e30199359ed09f1088b1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1cb9d1edb88c9a6545434305a22bfe5c3de7e2d3edbc75c2d3b81ccda30356be"}
{"qa_id": "uvo__0oYBU5SrAsc__object_world_motion_direction__00", "video_id": "uvo__0oYBU5SrAsc", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object B (brown dog) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "backward", "forward", "up"], "frame_paths": ["media/b2dcb99fe106fe023bc9/F0.jpg", "media/b2dcb99fe106fe023bc9/F1.jpg", "media/b2dcb99fe106fe023bc9/F2.jpg", "media/b2dcb99fe106fe023bc9/F3.jpg", "media/b2dcb99fe106fe023bc9/F4.jpg", "media/b2dcb99fe106fe023bc9/F5.jpg", "media/b2dcb99fe106fe023bc9/F6.jpg", "media/b2dcb99fe106fe023bc9/F7.jpg", "media/b2dcb99fe106fe023bc9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object B (brown dog) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5c4c3ec769d7999ff58bc21b9a216f3f0ea416d4ad5a69af0a0c647bab901fb"}
{"qa_id": "uvo__0pbMyICJUPw__camera_view_direction_change__00", "video_id": "uvo__0pbMyICJUPw", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/ec891172e48279bc93da/F0.jpg", "media/ec891172e48279bc93da/F1.jpg", "media/ec891172e48279bc93da/F2.jpg", "media/ec891172e48279bc93da/F3.jpg", "media/ec891172e48279bc93da/F4.jpg", "media/ec891172e48279bc93da/F5.jpg", "media/ec891172e48279bc93da/F6.jpg", "media/ec891172e48279bc93da/F7.jpg", "media/ec891172e48279bc93da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ef9d9350288380002a1215f19be1ed28c8850992f1d1a9e75527bf041ae9a88c"}
{"qa_id": "uvo__0pbMyICJUPw__object_world_motion_direction__01", "video_id": "uvo__0pbMyICJUPw", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (young boy with blond hair wearing a black t-shirt with a yellow batman logo) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/3c1c7330b9c0ee901758/F0.jpg", "media/3c1c7330b9c0ee901758/F1.jpg", "media/3c1c7330b9c0ee901758/F2.jpg", "media/3c1c7330b9c0ee901758/F3.jpg", "media/3c1c7330b9c0ee901758/F4.jpg", "media/3c1c7330b9c0ee901758/F5.jpg", "media/3c1c7330b9c0ee901758/F6.jpg", "media/3c1c7330b9c0ee901758/F7.jpg", "media/3c1c7330b9c0ee901758/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (young boy with blond hair wearing a black t-shirt with a yellow batman logo) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8caabb31eda478bb7f0c99461e31422baf32d082b457d211067ae3dcef316d7b"}
{"qa_id": "uvo__0pbMyICJUPw__object_travel_distance_comparison__02", "video_id": "uvo__0pbMyICJUPw", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (young boy with blond hair wearing a black t-shirt with a yellow batman logo) or B (young girl with a bob haircut wearing a pink patterned dress)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2e369995accdf2059792/F0.jpg", "media/2e369995accdf2059792/F1.jpg", "media/2e369995accdf2059792/F2.jpg", "media/2e369995accdf2059792/F3.jpg", "media/2e369995accdf2059792/F4.jpg", "media/2e369995accdf2059792/F5.jpg", "media/2e369995accdf2059792/F6.jpg", "media/2e369995accdf2059792/F7.jpg", "media/2e369995accdf2059792/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (young boy with blond hair wearing a black t-shirt with a yellow batman logo) or B (young girl with a bob haircut wearing a pink patterned dress)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "46d60b758fb669fffad6b49769c46ee7f5572a1cd31013617356f6c53044ae23"}
{"qa_id": "uvo__0s2mvbfzzyc__camera_translation_direction__00", "video_id": "uvo__0s2mvbfzzyc", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "down", "forward", "backward", "right"], "frame_paths": ["media/a07ba01979485472c190/F0.jpg", "media/a07ba01979485472c190/F1.jpg", "media/a07ba01979485472c190/F2.jpg", "media/a07ba01979485472c190/F3.jpg", "media/a07ba01979485472c190/F4.jpg", "media/a07ba01979485472c190/F5.jpg", "media/a07ba01979485472c190/F6.jpg", "media/a07ba01979485472c190/F7.jpg", "media/a07ba01979485472c190/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. down\nD. forward\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "45f0614750b40e8340b3cc572afee398484ae928f755015b4f89709f6f0ff7c4"}
{"qa_id": "uvo__0s2mvbfzzyc__camera_view_direction_change__01", "video_id": "uvo__0s2mvbfzzyc", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/a07ba01979485472c190/F0.jpg", "media/a07ba01979485472c190/F1.jpg", "media/a07ba01979485472c190/F2.jpg", "media/a07ba01979485472c190/F3.jpg", "media/a07ba01979485472c190/F4.jpg", "media/a07ba01979485472c190/F5.jpg", "media/a07ba01979485472c190/F6.jpg", "media/a07ba01979485472c190/F7.jpg", "media/a07ba01979485472c190/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "48b0dbe5136add4561fdd3af171b3ed657216b06d587528f3932aa08151751c1"}
{"qa_id": "uvo__0s2mvbfzzyc__camera_object_distance_change__02", "video_id": "uvo__0s2mvbfzzyc", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (person in dark clothing rappelling down a cliff face) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/24f43337f7495ecd3fc1/F0.jpg", "media/24f43337f7495ecd3fc1/F1.jpg", "media/24f43337f7495ecd3fc1/F2.jpg", "media/24f43337f7495ecd3fc1/F3.jpg", "media/24f43337f7495ecd3fc1/F4.jpg", "media/24f43337f7495ecd3fc1/F5.jpg", "media/24f43337f7495ecd3fc1/F6.jpg", "media/24f43337f7495ecd3fc1/F7.jpg", "media/24f43337f7495ecd3fc1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person in dark clothing rappelling down a cliff face) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "adc6bd0b5fc3238bbf0de25f5fd04364ab7903cf2f57697d632025c1a59bc7fd"}
{"qa_id": "uvo__0sC7Ovz0h20__camera_view_direction_change__00", "video_id": "uvo__0sC7Ovz0h20", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/5e4afff2cd72b99de378/F0.jpg", "media/5e4afff2cd72b99de378/F1.jpg", "media/5e4afff2cd72b99de378/F2.jpg", "media/5e4afff2cd72b99de378/F3.jpg", "media/5e4afff2cd72b99de378/F4.jpg", "media/5e4afff2cd72b99de378/F5.jpg", "media/5e4afff2cd72b99de378/F6.jpg", "media/5e4afff2cd72b99de378/F7.jpg", "media/5e4afff2cd72b99de378/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "689dae4aa0b4f962ab5c50120bf67e432403813d18160a815794f01d30ce4b3c"}
{"qa_id": "uvo__0sC7Ovz0h20__object_pair_distance_change__01", "video_id": "uvo__0sC7Ovz0h20", "task": "object_pair_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F6, did the 3D distance between object A (person in white shirt and dark pants) and object B (person in dark blue shirt and shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2657d14e35a527340388/F0.jpg", "media/2657d14e35a527340388/F1.jpg", "media/2657d14e35a527340388/F2.jpg", "media/2657d14e35a527340388/F3.jpg", "media/2657d14e35a527340388/F4.jpg", "media/2657d14e35a527340388/F5.jpg", "media/2657d14e35a527340388/F6.jpg", "media/2657d14e35a527340388/F7.jpg", "media/2657d14e35a527340388/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did the 3D distance between object A (person in white shirt and dark pants) and object B (person in dark blue shirt and shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f23173d378f4425c6f000a3285c09b87f99161fc3b6951d0ac90fa4cf9030a21"}
{"qa_id": "uvo__0sC7Ovz0h20__object_travel_distance_comparison__02", "video_id": "uvo__0sC7Ovz0h20", "task": "object_travel_distance_comparison", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and dark pants) or B (person in dark blue shirt and shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2657d14e35a527340388/F0.jpg", "media/2657d14e35a527340388/F1.jpg", "media/2657d14e35a527340388/F2.jpg", "media/2657d14e35a527340388/F3.jpg", "media/2657d14e35a527340388/F4.jpg", "media/2657d14e35a527340388/F5.jpg", "media/2657d14e35a527340388/F6.jpg", "media/2657d14e35a527340388/F7.jpg", "media/2657d14e35a527340388/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and dark pants) or B (person in dark blue shirt and shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "58f0126361647018a07f21f5320b3a53f1cfd1ef0245c3b8f8d04d726de7f71d"}
{"qa_id": "uvo__0sCN0p4ZgeA__camera_translation_direction__00", "video_id": "uvo__0sCN0p4ZgeA", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "right", "backward", "up", "left"], "frame_paths": ["media/5d86f8942332f41766d1/F0.jpg", "media/5d86f8942332f41766d1/F1.jpg", "media/5d86f8942332f41766d1/F2.jpg", "media/5d86f8942332f41766d1/F3.jpg", "media/5d86f8942332f41766d1/F4.jpg", "media/5d86f8942332f41766d1/F5.jpg", "media/5d86f8942332f41766d1/F6.jpg", "media/5d86f8942332f41766d1/F7.jpg", "media/5d86f8942332f41766d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. right\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "049a92050c7e9b19d1c34d3f0649322fc6518ae30134d2b9f42ab24829930084"}
{"qa_id": "uvo__0sCN0p4ZgeA__camera_view_direction_change__01", "video_id": "uvo__0sCN0p4ZgeA", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/5d86f8942332f41766d1/F0.jpg", "media/5d86f8942332f41766d1/F1.jpg", "media/5d86f8942332f41766d1/F2.jpg", "media/5d86f8942332f41766d1/F3.jpg", "media/5d86f8942332f41766d1/F4.jpg", "media/5d86f8942332f41766d1/F5.jpg", "media/5d86f8942332f41766d1/F6.jpg", "media/5d86f8942332f41766d1/F7.jpg", "media/5d86f8942332f41766d1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bc12d79511e7a4c2703fe469657077dfa5a1342963c60f9b378f06626186956d"}
{"qa_id": "uvo__0sCN0p4ZgeA__object_world_motion_direction__02", "video_id": "uvo__0sCN0p4ZgeA", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in camouflage uniform and black balaclava holding a prop gun) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/820e9dede1cf6029bdae/F0.jpg", "media/820e9dede1cf6029bdae/F1.jpg", "media/820e9dede1cf6029bdae/F2.jpg", "media/820e9dede1cf6029bdae/F3.jpg", "media/820e9dede1cf6029bdae/F4.jpg", "media/820e9dede1cf6029bdae/F5.jpg", "media/820e9dede1cf6029bdae/F6.jpg", "media/820e9dede1cf6029bdae/F7.jpg", "media/820e9dede1cf6029bdae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in camouflage uniform and black balaclava holding a prop gun) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c596f4c2574dd51417f325a916cb8057bfaef404915df2b7b7626ab911862aaa"}
{"qa_id": "uvo__0sCN0p4ZgeA__camera_object_distance_change__03", "video_id": "uvo__0sCN0p4ZgeA", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, did object A (person in camouflage uniform and black balaclava holding a prop gun) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/820e9dede1cf6029bdae/F0.jpg", "media/820e9dede1cf6029bdae/F1.jpg", "media/820e9dede1cf6029bdae/F2.jpg", "media/820e9dede1cf6029bdae/F3.jpg", "media/820e9dede1cf6029bdae/F4.jpg", "media/820e9dede1cf6029bdae/F5.jpg", "media/820e9dede1cf6029bdae/F6.jpg", "media/820e9dede1cf6029bdae/F7.jpg", "media/820e9dede1cf6029bdae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (person in camouflage uniform and black balaclava holding a prop gun) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "71eb00740e094871b3bcefe1d313af8379a7d0552d347fa6f85d721defa397dc"}
{"qa_id": "uvo__0sDydpFIu14__camera_translation_direction__00", "video_id": "uvo__0sDydpFIu14", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "right", "up", "forward"], "frame_paths": ["media/7efa22ecb0006bf41d2f/F0.jpg", "media/7efa22ecb0006bf41d2f/F1.jpg", "media/7efa22ecb0006bf41d2f/F2.jpg", "media/7efa22ecb0006bf41d2f/F3.jpg", "media/7efa22ecb0006bf41d2f/F4.jpg", "media/7efa22ecb0006bf41d2f/F5.jpg", "media/7efa22ecb0006bf41d2f/F6.jpg", "media/7efa22ecb0006bf41d2f/F7.jpg", "media/7efa22ecb0006bf41d2f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "08b4ff05aa5eead87fbf8035956f819557abb50b6ca8142327be271134ea50a6"}
{"qa_id": "uvo__0sDydpFIu14__camera_view_direction_change__01", "video_id": "uvo__0sDydpFIu14", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/7efa22ecb0006bf41d2f/F0.jpg", "media/7efa22ecb0006bf41d2f/F1.jpg", "media/7efa22ecb0006bf41d2f/F2.jpg", "media/7efa22ecb0006bf41d2f/F3.jpg", "media/7efa22ecb0006bf41d2f/F4.jpg", "media/7efa22ecb0006bf41d2f/F5.jpg", "media/7efa22ecb0006bf41d2f/F6.jpg", "media/7efa22ecb0006bf41d2f/F7.jpg", "media/7efa22ecb0006bf41d2f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "uvo__0sDydpFIu14__camera_object_distance_change__02", "video_id": "uvo__0sDydpFIu14", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, did object A (person in dark clothing climbing an ice wall) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c5530f61c9b9bab819bb/F0.jpg", "media/c5530f61c9b9bab819bb/F1.jpg", "media/c5530f61c9b9bab819bb/F2.jpg", "media/c5530f61c9b9bab819bb/F3.jpg", "media/c5530f61c9b9bab819bb/F4.jpg", "media/c5530f61c9b9bab819bb/F5.jpg", "media/c5530f61c9b9bab819bb/F6.jpg", "media/c5530f61c9b9bab819bb/F7.jpg", "media/c5530f61c9b9bab819bb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (person in dark clothing climbing an ice wall) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "86c213a0cde8afbe7ff51a60d93a15a43be979dd663d2da52ae1e4a7ddb2789b"}
{"qa_id": "uvo__0ssER51PqQc__camera_translation_direction__00", "video_id": "uvo__0ssER51PqQc", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "right", "forward", "up", "left"], "frame_paths": ["media/56b095416b095baa2d11/F0.jpg", "media/56b095416b095baa2d11/F1.jpg", "media/56b095416b095baa2d11/F2.jpg", "media/56b095416b095baa2d11/F3.jpg", "media/56b095416b095baa2d11/F4.jpg", "media/56b095416b095baa2d11/F5.jpg", "media/56b095416b095baa2d11/F6.jpg", "media/56b095416b095baa2d11/F7.jpg", "media/56b095416b095baa2d11/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. right\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "47e018eca7ba5a9d5f01dba5aaf8ff203bfda13578046368bd3acd5426e149de"}
{"qa_id": "uvo__0t-YfvZNdks__camera_view_direction_change__00", "video_id": "uvo__0t-YfvZNdks", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/0fed09b7864f8c3d65e4/F0.jpg", "media/0fed09b7864f8c3d65e4/F1.jpg", "media/0fed09b7864f8c3d65e4/F2.jpg", "media/0fed09b7864f8c3d65e4/F3.jpg", "media/0fed09b7864f8c3d65e4/F4.jpg", "media/0fed09b7864f8c3d65e4/F5.jpg", "media/0fed09b7864f8c3d65e4/F6.jpg", "media/0fed09b7864f8c3d65e4/F7.jpg", "media/0fed09b7864f8c3d65e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "76c9c3f5961d9d6f52825eb0d0e515757efff2935c5c5bdce255bc5e2350e01b"}
{"qa_id": "uvo__0t-YfvZNdks__camera_object_distance_change__01", "video_id": "uvo__0t-YfvZNdks", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (person wearing a blue and white striped shirt riding a camel) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b6a8c03c73ec3974c398/F0.jpg", "media/b6a8c03c73ec3974c398/F1.jpg", "media/b6a8c03c73ec3974c398/F2.jpg", "media/b6a8c03c73ec3974c398/F3.jpg", "media/b6a8c03c73ec3974c398/F4.jpg", "media/b6a8c03c73ec3974c398/F5.jpg", "media/b6a8c03c73ec3974c398/F6.jpg", "media/b6a8c03c73ec3974c398/F7.jpg", "media/b6a8c03c73ec3974c398/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person wearing a blue and white striped shirt riding a camel) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d7202eb7499d968488945a4bb8eb864f75fd31dbf90d70f70dabad1da39e62f5"}
{"qa_id": "uvo__0t-YfvZNdks__object_world_motion_direction__02", "video_id": "uvo__0t-YfvZNdks", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (person wearing a white shirt and dark pants walking beside a camel) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "right", "forward", "up"], "frame_paths": ["media/b1b023eac2b91f68b884/F0.jpg", "media/b1b023eac2b91f68b884/F1.jpg", "media/b1b023eac2b91f68b884/F2.jpg", "media/b1b023eac2b91f68b884/F3.jpg", "media/b1b023eac2b91f68b884/F4.jpg", "media/b1b023eac2b91f68b884/F5.jpg", "media/b1b023eac2b91f68b884/F6.jpg", "media/b1b023eac2b91f68b884/F7.jpg", "media/b1b023eac2b91f68b884/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (person wearing a white shirt and dark pants walking beside a camel) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e19b711eddf7697820e5b0e1207bfe906ecba28767eab24149d38d989033b00b"}
{"qa_id": "uvo__0t7kdrXsS5Q__object_pair_distance_change__00", "video_id": "uvo__0t7kdrXsS5Q", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F8, did the 3D distance between object A (person in dark clothing) and object B (person in red shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/01019004b78d4677dfab/F0.jpg", "media/01019004b78d4677dfab/F1.jpg", "media/01019004b78d4677dfab/F2.jpg", "media/01019004b78d4677dfab/F3.jpg", "media/01019004b78d4677dfab/F4.jpg", "media/01019004b78d4677dfab/F5.jpg", "media/01019004b78d4677dfab/F6.jpg", "media/01019004b78d4677dfab/F7.jpg", "media/01019004b78d4677dfab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person in dark clothing) and object B (person in red shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5b3bfa6ab306799cfac883a553c66ca8ee3201c26700ba3191c4bc485f4a940c"}
{"qa_id": "uvo__0t7kdrXsS5Q__object_travel_distance_comparison__01", "video_id": "uvo__0t7kdrXsS5Q", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in dark clothing) or B (person in red shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/01019004b78d4677dfab/F0.jpg", "media/01019004b78d4677dfab/F1.jpg", "media/01019004b78d4677dfab/F2.jpg", "media/01019004b78d4677dfab/F3.jpg", "media/01019004b78d4677dfab/F4.jpg", "media/01019004b78d4677dfab/F5.jpg", "media/01019004b78d4677dfab/F6.jpg", "media/01019004b78d4677dfab/F7.jpg", "media/01019004b78d4677dfab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in dark clothing) or B (person in red shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4c9afc3cea1b765ee244877ffac0b9a5fc275150f0c618f68f812f2dc029f522"}
{"qa_id": "uvo__0vXBP5tdNtg__camera_translation_direction__00", "video_id": "uvo__0vXBP5tdNtg", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "left", "forward", "down"], "frame_paths": ["media/0c756b16a83aeef6fa3e/F0.jpg", "media/0c756b16a83aeef6fa3e/F1.jpg", "media/0c756b16a83aeef6fa3e/F2.jpg", "media/0c756b16a83aeef6fa3e/F3.jpg", "media/0c756b16a83aeef6fa3e/F4.jpg", "media/0c756b16a83aeef6fa3e/F5.jpg", "media/0c756b16a83aeef6fa3e/F6.jpg", "media/0c756b16a83aeef6fa3e/F7.jpg", "media/0c756b16a83aeef6fa3e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7b6ba8cd3dd3f98498f8fb3cfd335d63f61d2e58f50386317329cd297a89a367"}
{"qa_id": "uvo__0vfPZXdnWnU__object_world_motion_direction__00", "video_id": "uvo__0vfPZXdnWnU", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and orange shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "backward", "up", "forward", "left"], "frame_paths": ["media/f5bc9750abc35d609cdd/F0.jpg", "media/f5bc9750abc35d609cdd/F1.jpg", "media/f5bc9750abc35d609cdd/F2.jpg", "media/f5bc9750abc35d609cdd/F3.jpg", "media/f5bc9750abc35d609cdd/F4.jpg", "media/f5bc9750abc35d609cdd/F5.jpg", "media/f5bc9750abc35d609cdd/F6.jpg", "media/f5bc9750abc35d609cdd/F7.jpg", "media/f5bc9750abc35d609cdd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and orange shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. backward\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "09869ca4ff3280155c7109fcb306e2557ef8d019c09f376473f06dd22daea76f"}
{"qa_id": "uvo__0vlR8sTEsXE__camera_view_direction_change__00", "video_id": "uvo__0vlR8sTEsXE", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/4059a7b2128114baac76/F0.jpg", "media/4059a7b2128114baac76/F1.jpg", "media/4059a7b2128114baac76/F2.jpg", "media/4059a7b2128114baac76/F3.jpg", "media/4059a7b2128114baac76/F4.jpg", "media/4059a7b2128114baac76/F5.jpg", "media/4059a7b2128114baac76/F6.jpg", "media/4059a7b2128114baac76/F7.jpg", "media/4059a7b2128114baac76/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d8e14f91696e80ae6ee9241b01be6fd3ed02474c4407b05076297a377d3021fe"}
{"qa_id": "uvo__0wXbXtvFpUI__camera_translation_direction__00", "video_id": "uvo__0wXbXtvFpUI", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/170ee3b9b52aaf93dfcf/F0.jpg", "media/170ee3b9b52aaf93dfcf/F1.jpg", "media/170ee3b9b52aaf93dfcf/F2.jpg", "media/170ee3b9b52aaf93dfcf/F3.jpg", "media/170ee3b9b52aaf93dfcf/F4.jpg", "media/170ee3b9b52aaf93dfcf/F5.jpg", "media/170ee3b9b52aaf93dfcf/F6.jpg", "media/170ee3b9b52aaf93dfcf/F7.jpg", "media/170ee3b9b52aaf93dfcf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "90c1fff678d5e33b868625f6eccad1f063303bf08a67dd7284d1987beb30c9be"}
{"qa_id": "uvo__0wXbXtvFpUI__camera_view_direction_change__01", "video_id": "uvo__0wXbXtvFpUI", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/170ee3b9b52aaf93dfcf/F0.jpg", "media/170ee3b9b52aaf93dfcf/F1.jpg", "media/170ee3b9b52aaf93dfcf/F2.jpg", "media/170ee3b9b52aaf93dfcf/F3.jpg", "media/170ee3b9b52aaf93dfcf/F4.jpg", "media/170ee3b9b52aaf93dfcf/F5.jpg", "media/170ee3b9b52aaf93dfcf/F6.jpg", "media/170ee3b9b52aaf93dfcf/F7.jpg", "media/170ee3b9b52aaf93dfcf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "uvo__0wXbXtvFpUI__object_world_motion_direction__02", "video_id": "uvo__0wXbXtvFpUI", "task": "object_world_motion_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (child in red shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "up", "left", "down"], "frame_paths": ["media/50c0864a0523b8ada701/F0.jpg", "media/50c0864a0523b8ada701/F1.jpg", "media/50c0864a0523b8ada701/F2.jpg", "media/50c0864a0523b8ada701/F3.jpg", "media/50c0864a0523b8ada701/F4.jpg", "media/50c0864a0523b8ada701/F5.jpg", "media/50c0864a0523b8ada701/F6.jpg", "media/50c0864a0523b8ada701/F7.jpg", "media/50c0864a0523b8ada701/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (child in red shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. up\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4dce7da6b5f3449602eec843e4a6f58af147750f7b5b97159debf6b3d0300d6"}
{"qa_id": "uvo__0w_ShCl1xvg__camera_view_direction_change__00", "video_id": "uvo__0w_ShCl1xvg", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/b049fad5aea2dd2050aa/F0.jpg", "media/b049fad5aea2dd2050aa/F1.jpg", "media/b049fad5aea2dd2050aa/F2.jpg", "media/b049fad5aea2dd2050aa/F3.jpg", "media/b049fad5aea2dd2050aa/F4.jpg", "media/b049fad5aea2dd2050aa/F5.jpg", "media/b049fad5aea2dd2050aa/F6.jpg", "media/b049fad5aea2dd2050aa/F7.jpg", "media/b049fad5aea2dd2050aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6e894bc1bf0ae5cd4ff6634e46928a156052b6c353aca1b723fe2310a012edcd"}
{"qa_id": "uvo__0w_ShCl1xvg__object_world_motion_direction__01", "video_id": "uvo__0w_ShCl1xvg", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (young girl with a ponytail wearing a white long-sleeved shirt with a cartoon graphic and pink pants) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "down", "forward", "backward"], "frame_paths": ["media/188a9f11b18e8e5dd482/F0.jpg", "media/188a9f11b18e8e5dd482/F1.jpg", "media/188a9f11b18e8e5dd482/F2.jpg", "media/188a9f11b18e8e5dd482/F3.jpg", "media/188a9f11b18e8e5dd482/F4.jpg", "media/188a9f11b18e8e5dd482/F5.jpg", "media/188a9f11b18e8e5dd482/F6.jpg", "media/188a9f11b18e8e5dd482/F7.jpg", "media/188a9f11b18e8e5dd482/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (young girl with a ponytail wearing a white long-sleeved shirt with a cartoon graphic and pink pants) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. down\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "28c8f9501c7db4092861c26ab808689ffa43a78caa2a3bbadc85d9733afd72e8"}
{"qa_id": "uvo__0w_yy292WWM__object_world_motion_direction__00", "video_id": "uvo__0w_yy292WWM", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (shirtless man in dark shorts) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "left", "up", "backward"], "frame_paths": ["media/a5531da9d8f767a93e22/F0.jpg", "media/a5531da9d8f767a93e22/F1.jpg", "media/a5531da9d8f767a93e22/F2.jpg", "media/a5531da9d8f767a93e22/F3.jpg", "media/a5531da9d8f767a93e22/F4.jpg", "media/a5531da9d8f767a93e22/F5.jpg", "media/a5531da9d8f767a93e22/F6.jpg", "media/a5531da9d8f767a93e22/F7.jpg", "media/a5531da9d8f767a93e22/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (shirtless man in dark shorts) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0268fb690acbba29084157e5e2ff7769177f46aaba8f0f2ac986a7d31de7d3fa"}
{"qa_id": "uvo__0xOnw8cvcTQ__camera_view_direction_change__00", "video_id": "uvo__0xOnw8cvcTQ", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/d17eb7ded76b0f0318e7/F0.jpg", "media/d17eb7ded76b0f0318e7/F1.jpg", "media/d17eb7ded76b0f0318e7/F2.jpg", "media/d17eb7ded76b0f0318e7/F3.jpg", "media/d17eb7ded76b0f0318e7/F4.jpg", "media/d17eb7ded76b0f0318e7/F5.jpg", "media/d17eb7ded76b0f0318e7/F6.jpg", "media/d17eb7ded76b0f0318e7/F7.jpg", "media/d17eb7ded76b0f0318e7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__0xOnw8cvcTQ__object_pair_distance_change__01", "video_id": "uvo__0xOnw8cvcTQ", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F2, did the 3D distance between object A (person walking on the right side of the road) and object B (person in the middle distance near a blue shed) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/93700aca218ce4c55d82/F0.jpg", "media/93700aca218ce4c55d82/F1.jpg", "media/93700aca218ce4c55d82/F2.jpg", "media/93700aca218ce4c55d82/F3.jpg", "media/93700aca218ce4c55d82/F4.jpg", "media/93700aca218ce4c55d82/F5.jpg", "media/93700aca218ce4c55d82/F6.jpg", "media/93700aca218ce4c55d82/F7.jpg", "media/93700aca218ce4c55d82/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (person walking on the right side of the road) and object B (person in the middle distance near a blue shed) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "662ab0823b52432aa38bef6c1c385b0e5e2c97b7fb90e374e665226163eb3aed"}
{"qa_id": "uvo__0xOnw8cvcTQ__object_travel_distance_comparison__02", "video_id": "uvo__0xOnw8cvcTQ", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person walking on the right side of the road) or B (person in the middle distance near a blue shed)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/93700aca218ce4c55d82/F0.jpg", "media/93700aca218ce4c55d82/F1.jpg", "media/93700aca218ce4c55d82/F2.jpg", "media/93700aca218ce4c55d82/F3.jpg", "media/93700aca218ce4c55d82/F4.jpg", "media/93700aca218ce4c55d82/F5.jpg", "media/93700aca218ce4c55d82/F6.jpg", "media/93700aca218ce4c55d82/F7.jpg", "media/93700aca218ce4c55d82/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person walking on the right side of the road) or B (person in the middle distance near a blue shed)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "39b3648fa8d5b85fe441901379e3ebb79d316a3a5500cd27c7c087ec33d03981"}
{"qa_id": "uvo__0zCzhU85ESU__camera_view_direction_change__00", "video_id": "uvo__0zCzhU85ESU", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "down", "right", "left"], "frame_paths": ["media/7a52c46c849bfc4da6f6/F0.jpg", "media/7a52c46c849bfc4da6f6/F1.jpg", "media/7a52c46c849bfc4da6f6/F2.jpg", "media/7a52c46c849bfc4da6f6/F3.jpg", "media/7a52c46c849bfc4da6f6/F4.jpg", "media/7a52c46c849bfc4da6f6/F5.jpg", "media/7a52c46c849bfc4da6f6/F6.jpg", "media/7a52c46c849bfc4da6f6/F7.jpg", "media/7a52c46c849bfc4da6f6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. down\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8e1e3dd0ead9cad96fcc89538b61d76b96adf4a4547c8986e25ef300851b302"}
{"qa_id": "uvo__0zu0S6ttK3w__camera_view_direction_change__00", "video_id": "uvo__0zu0S6ttK3w", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/cf6c3bec93c023c2486c/F0.jpg", "media/cf6c3bec93c023c2486c/F1.jpg", "media/cf6c3bec93c023c2486c/F2.jpg", "media/cf6c3bec93c023c2486c/F3.jpg", "media/cf6c3bec93c023c2486c/F4.jpg", "media/cf6c3bec93c023c2486c/F5.jpg", "media/cf6c3bec93c023c2486c/F6.jpg", "media/cf6c3bec93c023c2486c/F7.jpg", "media/cf6c3bec93c023c2486c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da89b6fca02dc341645573855006cd191b5820895671cb7af57de3a5554a1807"}
{"qa_id": "uvo__1-41f7JtnW4__camera_translation_direction__00", "video_id": "uvo__1-41f7JtnW4", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "backward", "down", "left"], "frame_paths": ["media/60d0c364e5ef351f5b80/F0.jpg", "media/60d0c364e5ef351f5b80/F1.jpg", "media/60d0c364e5ef351f5b80/F2.jpg", "media/60d0c364e5ef351f5b80/F3.jpg", "media/60d0c364e5ef351f5b80/F4.jpg", "media/60d0c364e5ef351f5b80/F5.jpg", "media/60d0c364e5ef351f5b80/F6.jpg", "media/60d0c364e5ef351f5b80/F7.jpg", "media/60d0c364e5ef351f5b80/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fdc304ac83f011c013321abe11e01af99ec104ea7fa82f4862f8e0a1704fa2e5"}
{"qa_id": "uvo__1-41f7JtnW4__camera_view_direction_change__01", "video_id": "uvo__1-41f7JtnW4", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/60d0c364e5ef351f5b80/F0.jpg", "media/60d0c364e5ef351f5b80/F1.jpg", "media/60d0c364e5ef351f5b80/F2.jpg", "media/60d0c364e5ef351f5b80/F3.jpg", "media/60d0c364e5ef351f5b80/F4.jpg", "media/60d0c364e5ef351f5b80/F5.jpg", "media/60d0c364e5ef351f5b80/F6.jpg", "media/60d0c364e5ef351f5b80/F7.jpg", "media/60d0c364e5ef351f5b80/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5037fe4e211d21b5c4b173973ab9623df2497eee16f8e4379ca89dbde8e71582"}
{"qa_id": "uvo__1-HzoNEwKO4__object_world_motion_direction__00", "video_id": "uvo__1-HzoNEwKO4", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (adult wearing a dark plaid shirt sitting behind a child on a sled) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "backward", "down", "up"], "frame_paths": ["media/508060d25e5e592f82ff/F0.jpg", "media/508060d25e5e592f82ff/F1.jpg", "media/508060d25e5e592f82ff/F2.jpg", "media/508060d25e5e592f82ff/F3.jpg", "media/508060d25e5e592f82ff/F4.jpg", "media/508060d25e5e592f82ff/F5.jpg", "media/508060d25e5e592f82ff/F6.jpg", "media/508060d25e5e592f82ff/F7.jpg", "media/508060d25e5e592f82ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (adult wearing a dark plaid shirt sitting behind a child on a sled) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d906b466b48dc3df1ea35ac767c9b592ae8b6de276ea925119a8b6ee34e383b8"}
{"qa_id": "uvo__1-HzoNEwKO4__camera_object_distance_change__01", "video_id": "uvo__1-HzoNEwKO4", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (adult wearing a dark plaid shirt sitting behind a child on a sled) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/508060d25e5e592f82ff/F0.jpg", "media/508060d25e5e592f82ff/F1.jpg", "media/508060d25e5e592f82ff/F2.jpg", "media/508060d25e5e592f82ff/F3.jpg", "media/508060d25e5e592f82ff/F4.jpg", "media/508060d25e5e592f82ff/F5.jpg", "media/508060d25e5e592f82ff/F6.jpg", "media/508060d25e5e592f82ff/F7.jpg", "media/508060d25e5e592f82ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (adult wearing a dark plaid shirt sitting behind a child on a sled) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e3aa888ee93fdb81c3c2c48f2d8c2e9d3ea0eb90020e8a6740c166087db082c9"}
{"qa_id": "uvo__1-HzoNEwKO4__object_world_motion_direction__02", "video_id": "uvo__1-HzoNEwKO4", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (child wearing an orange jacket sitting on a sled) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "left", "right"], "frame_paths": ["media/fb1dec8f47af0b935d75/F0.jpg", "media/fb1dec8f47af0b935d75/F1.jpg", "media/fb1dec8f47af0b935d75/F2.jpg", "media/fb1dec8f47af0b935d75/F3.jpg", "media/fb1dec8f47af0b935d75/F4.jpg", "media/fb1dec8f47af0b935d75/F5.jpg", "media/fb1dec8f47af0b935d75/F6.jpg", "media/fb1dec8f47af0b935d75/F7.jpg", "media/fb1dec8f47af0b935d75/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (child wearing an orange jacket sitting on a sled) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1992ba135530124bdd660c529b3dda56055f567574bc0e4d260d7d89c6afb151"}
{"qa_id": "uvo__1-HzoNEwKO4__camera_object_distance_change__03", "video_id": "uvo__1-HzoNEwKO4", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object B (child wearing an orange jacket sitting on a sled) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/fb1dec8f47af0b935d75/F0.jpg", "media/fb1dec8f47af0b935d75/F1.jpg", "media/fb1dec8f47af0b935d75/F2.jpg", "media/fb1dec8f47af0b935d75/F3.jpg", "media/fb1dec8f47af0b935d75/F4.jpg", "media/fb1dec8f47af0b935d75/F5.jpg", "media/fb1dec8f47af0b935d75/F6.jpg", "media/fb1dec8f47af0b935d75/F7.jpg", "media/fb1dec8f47af0b935d75/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object B (child wearing an orange jacket sitting on a sled) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3ddfdc6b18320e14febd78ef0bfdfa79faa378c40c78d946bd4a5801b46a64ab"}
{"qa_id": "uvo__1-Y5z-CE_B0__camera_translation_direction__00", "video_id": "uvo__1-Y5z-CE_B0", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "forward", "right", "backward"], "frame_paths": ["media/43e6e53b70b752df5af6/F0.jpg", "media/43e6e53b70b752df5af6/F1.jpg", "media/43e6e53b70b752df5af6/F2.jpg", "media/43e6e53b70b752df5af6/F3.jpg", "media/43e6e53b70b752df5af6/F4.jpg", "media/43e6e53b70b752df5af6/F5.jpg", "media/43e6e53b70b752df5af6/F6.jpg", "media/43e6e53b70b752df5af6/F7.jpg", "media/43e6e53b70b752df5af6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a0dd077db69a0040b8966ea0b37f53944f097f04fc81ba462a05be5384838b77"}
{"qa_id": "uvo__1-yzOGhaXD8__camera_object_distance_change__00", "video_id": "uvo__1-yzOGhaXD8", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object B (person in red jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/1641f3c82d3bbec80a92/F0.jpg", "media/1641f3c82d3bbec80a92/F1.jpg", "media/1641f3c82d3bbec80a92/F2.jpg", "media/1641f3c82d3bbec80a92/F3.jpg", "media/1641f3c82d3bbec80a92/F4.jpg", "media/1641f3c82d3bbec80a92/F5.jpg", "media/1641f3c82d3bbec80a92/F6.jpg", "media/1641f3c82d3bbec80a92/F7.jpg", "media/1641f3c82d3bbec80a92/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (person in red jacket and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e82bb677bc5772f6f277e5edebfc876c402d31c3bbb4e008a7350ffb105fbcb"}
{"qa_id": "uvo__1-yzOGhaXD8__object_travel_distance_comparison__01", "video_id": "uvo__1-yzOGhaXD8", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange jacket pulling a sled) or B (person in red jacket and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f76ddf7f407743a59415/F0.jpg", "media/f76ddf7f407743a59415/F1.jpg", "media/f76ddf7f407743a59415/F2.jpg", "media/f76ddf7f407743a59415/F3.jpg", "media/f76ddf7f407743a59415/F4.jpg", "media/f76ddf7f407743a59415/F5.jpg", "media/f76ddf7f407743a59415/F6.jpg", "media/f76ddf7f407743a59415/F7.jpg", "media/f76ddf7f407743a59415/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange jacket pulling a sled) or B (person in red jacket and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc721b8a4bf3e1102650dbe1a63dea05381e305b74f0679c929f17dfa1d8ddb6"}
{"qa_id": "uvo__11IUeetn2iM__camera_view_direction_change__00", "video_id": "uvo__11IUeetn2iM", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/10ab6f5b4b942466bb98/F0.jpg", "media/10ab6f5b4b942466bb98/F1.jpg", "media/10ab6f5b4b942466bb98/F2.jpg", "media/10ab6f5b4b942466bb98/F3.jpg", "media/10ab6f5b4b942466bb98/F4.jpg", "media/10ab6f5b4b942466bb98/F5.jpg", "media/10ab6f5b4b942466bb98/F6.jpg", "media/10ab6f5b4b942466bb98/F7.jpg", "media/10ab6f5b4b942466bb98/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b4bb7ec21e095c5c46cce0347f69602ee73fee0909fed7189c01b64cbf085997"}
{"qa_id": "uvo__11i9oITK3_8__camera_view_direction_change__00", "video_id": "uvo__11i9oITK3_8", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/6ae11df1e0dec5a94847/F0.jpg", "media/6ae11df1e0dec5a94847/F1.jpg", "media/6ae11df1e0dec5a94847/F2.jpg", "media/6ae11df1e0dec5a94847/F3.jpg", "media/6ae11df1e0dec5a94847/F4.jpg", "media/6ae11df1e0dec5a94847/F5.jpg", "media/6ae11df1e0dec5a94847/F6.jpg", "media/6ae11df1e0dec5a94847/F7.jpg", "media/6ae11df1e0dec5a94847/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "613fd6506cc60935fcb5f5397bf7795f6b9c1aac25e52452d8d03758d13c660d"}
{"qa_id": "uvo__12nT-AHwIdM__camera_translation_direction__00", "video_id": "uvo__12nT-AHwIdM", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/2a69b698d65f8a63caa9/F0.jpg", "media/2a69b698d65f8a63caa9/F1.jpg", "media/2a69b698d65f8a63caa9/F2.jpg", "media/2a69b698d65f8a63caa9/F3.jpg", "media/2a69b698d65f8a63caa9/F4.jpg", "media/2a69b698d65f8a63caa9/F5.jpg", "media/2a69b698d65f8a63caa9/F6.jpg", "media/2a69b698d65f8a63caa9/F7.jpg", "media/2a69b698d65f8a63caa9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "539726ef021f37ef8f02bac0477fe1a9d93560afe3bbc0e580008be882053aea"}
{"qa_id": "uvo__12nT-AHwIdM__camera_view_direction_change__01", "video_id": "uvo__12nT-AHwIdM", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/2a69b698d65f8a63caa9/F0.jpg", "media/2a69b698d65f8a63caa9/F1.jpg", "media/2a69b698d65f8a63caa9/F2.jpg", "media/2a69b698d65f8a63caa9/F3.jpg", "media/2a69b698d65f8a63caa9/F4.jpg", "media/2a69b698d65f8a63caa9/F5.jpg", "media/2a69b698d65f8a63caa9/F6.jpg", "media/2a69b698d65f8a63caa9/F7.jpg", "media/2a69b698d65f8a63caa9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2fa94d91120fdc4f8f000e85c881bae7620fb4e698ec7af609627304d60dfe97"}
{"qa_id": "uvo__12nT-AHwIdM__object_world_motion_direction__02", "video_id": "uvo__12nT-AHwIdM", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in red ski suit) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "up", "backward", "right", "down"], "frame_paths": ["media/57a31502f50be1140d62/F0.jpg", "media/57a31502f50be1140d62/F1.jpg", "media/57a31502f50be1140d62/F2.jpg", "media/57a31502f50be1140d62/F3.jpg", "media/57a31502f50be1140d62/F4.jpg", "media/57a31502f50be1140d62/F5.jpg", "media/57a31502f50be1140d62/F6.jpg", "media/57a31502f50be1140d62/F7.jpg", "media/57a31502f50be1140d62/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in red ski suit) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. up\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3c6d091194a2deaffe7117f63516b580a2ffc70d7e492270311a3d40ac0992f"}
{"qa_id": "uvo__12vWZg5BdG4__camera_translation_direction__00", "video_id": "uvo__12vWZg5BdG4", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "up", "forward", "backward", "left", "right"], "frame_paths": ["media/ba244fff430b165100e4/F0.jpg", "media/ba244fff430b165100e4/F1.jpg", "media/ba244fff430b165100e4/F2.jpg", "media/ba244fff430b165100e4/F3.jpg", "media/ba244fff430b165100e4/F4.jpg", "media/ba244fff430b165100e4/F5.jpg", "media/ba244fff430b165100e4/F6.jpg", "media/ba244fff430b165100e4/F7.jpg", "media/ba244fff430b165100e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. forward\nD. backward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "14172e0f2a0e1981dbd0f18102edbeaf7a16523cf13f47a6f44d3f9e77a9c2a3"}
{"qa_id": "uvo__13sohDAPS0g__object_world_motion_direction__00", "video_id": "uvo__13sohDAPS0g", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in black shirt with white logo) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "right", "left", "forward"], "frame_paths": ["media/91422b0d1610416c5566/F0.jpg", "media/91422b0d1610416c5566/F1.jpg", "media/91422b0d1610416c5566/F2.jpg", "media/91422b0d1610416c5566/F3.jpg", "media/91422b0d1610416c5566/F4.jpg", "media/91422b0d1610416c5566/F5.jpg", "media/91422b0d1610416c5566/F6.jpg", "media/91422b0d1610416c5566/F7.jpg", "media/91422b0d1610416c5566/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in black shirt with white logo) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "057e3b222bfff4f022e5a26deabc37edd4b0c76d2a9a7332e6be144d55c524e0"}
{"qa_id": "uvo__14tjqjjjAEI__camera_translation_direction__00", "video_id": "uvo__14tjqjjjAEI", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "forward", "up", "backward"], "frame_paths": ["media/ca021ac283fafc240782/F0.jpg", "media/ca021ac283fafc240782/F1.jpg", "media/ca021ac283fafc240782/F2.jpg", "media/ca021ac283fafc240782/F3.jpg", "media/ca021ac283fafc240782/F4.jpg", "media/ca021ac283fafc240782/F5.jpg", "media/ca021ac283fafc240782/F6.jpg", "media/ca021ac283fafc240782/F7.jpg", "media/ca021ac283fafc240782/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. forward\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0a4ee66ff4e9dcacf3497a628278d5adef54ac4f22f06527c11a36f6e6f63e0f"}
{"qa_id": "uvo__14tjqjjjAEI__object_pair_distance_change__01", "video_id": "uvo__14tjqjjjAEI", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (person standing on a dock in the background) and object B (water skier in a black wetsuit holding a red and yellow object) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b578933c7bc8c478d539/F0.jpg", "media/b578933c7bc8c478d539/F1.jpg", "media/b578933c7bc8c478d539/F2.jpg", "media/b578933c7bc8c478d539/F3.jpg", "media/b578933c7bc8c478d539/F4.jpg", "media/b578933c7bc8c478d539/F5.jpg", "media/b578933c7bc8c478d539/F6.jpg", "media/b578933c7bc8c478d539/F7.jpg", "media/b578933c7bc8c478d539/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person standing on a dock in the background) and object B (water skier in a black wetsuit holding a red and yellow object) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e59dd99e784e10f55b9f059d7e5d4ebc2f6bde89d8df3c5004cbfeb5f5512e6d"}
{"qa_id": "uvo__14tjqjjjAEI__object_travel_distance_comparison__02", "video_id": "uvo__14tjqjjjAEI", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person standing on a dock in the background) or B (water skier in a black wetsuit holding a red and yellow object)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/b578933c7bc8c478d539/F0.jpg", "media/b578933c7bc8c478d539/F1.jpg", "media/b578933c7bc8c478d539/F2.jpg", "media/b578933c7bc8c478d539/F3.jpg", "media/b578933c7bc8c478d539/F4.jpg", "media/b578933c7bc8c478d539/F5.jpg", "media/b578933c7bc8c478d539/F6.jpg", "media/b578933c7bc8c478d539/F7.jpg", "media/b578933c7bc8c478d539/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person standing on a dock in the background) or B (water skier in a black wetsuit holding a red and yellow object)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0bf5ffd0596e4dd1859f75371c8b85dc94aa33ca03d9b73ee36e4486caace89f"}
{"qa_id": "uvo__15JRTW-QAYA__camera_object_distance_change__00", "video_id": "uvo__15JRTW-QAYA", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (person in dark clothing sitting on a camel) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/39b635f57a928de49cc3/F0.jpg", "media/39b635f57a928de49cc3/F1.jpg", "media/39b635f57a928de49cc3/F2.jpg", "media/39b635f57a928de49cc3/F3.jpg", "media/39b635f57a928de49cc3/F4.jpg", "media/39b635f57a928de49cc3/F5.jpg", "media/39b635f57a928de49cc3/F6.jpg", "media/39b635f57a928de49cc3/F7.jpg", "media/39b635f57a928de49cc3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (person in dark clothing sitting on a camel) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6d7b652e8386890875c06824323c6a53af9389d07d9b546eaac0e247a73c36f8"}
{"qa_id": "uvo__15JRTW-QAYA__camera_object_distance_change__01", "video_id": "uvo__15JRTW-QAYA", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object B (person in green shirt sitting on a camel) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ea16507d9244ae30be2a/F0.jpg", "media/ea16507d9244ae30be2a/F1.jpg", "media/ea16507d9244ae30be2a/F2.jpg", "media/ea16507d9244ae30be2a/F3.jpg", "media/ea16507d9244ae30be2a/F4.jpg", "media/ea16507d9244ae30be2a/F5.jpg", "media/ea16507d9244ae30be2a/F6.jpg", "media/ea16507d9244ae30be2a/F7.jpg", "media/ea16507d9244ae30be2a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object B (person in green shirt sitting on a camel) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e52eba9c93e1210897707414e4ea4010e669305d4215956346cf0b8c91a94cc5"}
{"qa_id": "uvo__15LSh_XPils__object_world_motion_direction__00", "video_id": "uvo__15LSh_XPils", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (orange and blue paraglider canopy) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "right", "left", "down"], "frame_paths": ["media/6ffecb83abc7f9cd0d99/F0.jpg", "media/6ffecb83abc7f9cd0d99/F1.jpg", "media/6ffecb83abc7f9cd0d99/F2.jpg", "media/6ffecb83abc7f9cd0d99/F3.jpg", "media/6ffecb83abc7f9cd0d99/F4.jpg", "media/6ffecb83abc7f9cd0d99/F5.jpg", "media/6ffecb83abc7f9cd0d99/F6.jpg", "media/6ffecb83abc7f9cd0d99/F7.jpg", "media/6ffecb83abc7f9cd0d99/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (orange and blue paraglider canopy) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0dc8673fd36667bc639805511d77cbd07b0a3a40d51cabac68d6ad461f3f3c06"}
{"qa_id": "uvo__15SqU9wyekc__camera_view_direction_change__00", "video_id": "uvo__15SqU9wyekc", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/311018ea544f2e75caa6/F0.jpg", "media/311018ea544f2e75caa6/F1.jpg", "media/311018ea544f2e75caa6/F2.jpg", "media/311018ea544f2e75caa6/F3.jpg", "media/311018ea544f2e75caa6/F4.jpg", "media/311018ea544f2e75caa6/F5.jpg", "media/311018ea544f2e75caa6/F6.jpg", "media/311018ea544f2e75caa6/F7.jpg", "media/311018ea544f2e75caa6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "199105120d5aae11924ed2a33bf9bb468e8c513c19f24ed56588c61b3bf1f97e"}
{"qa_id": "uvo__161EH1x7Uxc__camera_translation_direction__00", "video_id": "uvo__161EH1x7Uxc", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "backward", "up"], "frame_paths": ["media/f6db9e67211c515d499e/F0.jpg", "media/f6db9e67211c515d499e/F1.jpg", "media/f6db9e67211c515d499e/F2.jpg", "media/f6db9e67211c515d499e/F3.jpg", "media/f6db9e67211c515d499e/F4.jpg", "media/f6db9e67211c515d499e/F5.jpg", "media/f6db9e67211c515d499e/F6.jpg", "media/f6db9e67211c515d499e/F7.jpg", "media/f6db9e67211c515d499e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "edd7999b47a2c7218d3f2d05a24f0686665701d2f40bcfc00515057e3bc4d0b7"}
{"qa_id": "uvo__161EH1x7Uxc__object_world_motion_direction__01", "video_id": "uvo__161EH1x7Uxc", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person standing on the side of the road) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "left", "up", "backward", "down"], "frame_paths": ["media/1b28fdd7b0c3810e7f21/F0.jpg", "media/1b28fdd7b0c3810e7f21/F1.jpg", "media/1b28fdd7b0c3810e7f21/F2.jpg", "media/1b28fdd7b0c3810e7f21/F3.jpg", "media/1b28fdd7b0c3810e7f21/F4.jpg", "media/1b28fdd7b0c3810e7f21/F5.jpg", "media/1b28fdd7b0c3810e7f21/F6.jpg", "media/1b28fdd7b0c3810e7f21/F7.jpg", "media/1b28fdd7b0c3810e7f21/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person standing on the side of the road) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. left\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e7ebf8e2f534c52aff5bc4f4767c771d0b40b54bf6724e30e1d0c500752e8479"}
{"qa_id": "uvo__168HoJooFQA__camera_view_direction_change__00", "video_id": "uvo__168HoJooFQA", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/a6f9b9d057d0d736334d/F0.jpg", "media/a6f9b9d057d0d736334d/F1.jpg", "media/a6f9b9d057d0d736334d/F2.jpg", "media/a6f9b9d057d0d736334d/F3.jpg", "media/a6f9b9d057d0d736334d/F4.jpg", "media/a6f9b9d057d0d736334d/F5.jpg", "media/a6f9b9d057d0d736334d/F6.jpg", "media/a6f9b9d057d0d736334d/F7.jpg", "media/a6f9b9d057d0d736334d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "689dae4aa0b4f962ab5c50120bf67e432403813d18160a815794f01d30ce4b3c"}
{"qa_id": "uvo__16a65wnUux0__object_world_motion_direction__00", "video_id": "uvo__16a65wnUux0", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (person in beige coat and red hat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "backward", "up", "forward"], "frame_paths": ["media/a895a0ed1307d95404ec/F0.jpg", "media/a895a0ed1307d95404ec/F1.jpg", "media/a895a0ed1307d95404ec/F2.jpg", "media/a895a0ed1307d95404ec/F3.jpg", "media/a895a0ed1307d95404ec/F4.jpg", "media/a895a0ed1307d95404ec/F5.jpg", "media/a895a0ed1307d95404ec/F6.jpg", "media/a895a0ed1307d95404ec/F7.jpg", "media/a895a0ed1307d95404ec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (person in beige coat and red hat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebb0a595819f67b5a2746d6394c5a3df80126c3ecb6965b4a4732fe2e64eb336"}
{"qa_id": "uvo__16a65wnUux0__object_pair_distance_change__01", "video_id": "uvo__16a65wnUux0", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (small orange dog) and object B (person in beige coat and red hat) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/5a343ec188d0194c41c4/F0.jpg", "media/5a343ec188d0194c41c4/F1.jpg", "media/5a343ec188d0194c41c4/F2.jpg", "media/5a343ec188d0194c41c4/F3.jpg", "media/5a343ec188d0194c41c4/F4.jpg", "media/5a343ec188d0194c41c4/F5.jpg", "media/5a343ec188d0194c41c4/F6.jpg", "media/5a343ec188d0194c41c4/F7.jpg", "media/5a343ec188d0194c41c4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (small orange dog) and object B (person in beige coat and red hat) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "77e20aefd92538f75445ee02c076b09152f34938f221ad51c94c40704cfcad7e"}
{"qa_id": "uvo__16dDTSncELA__camera_translation_direction__00", "video_id": "uvo__16dDTSncELA", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "right", "down", "up"], "frame_paths": ["media/5aaecd8df9286f4d1453/F0.jpg", "media/5aaecd8df9286f4d1453/F1.jpg", "media/5aaecd8df9286f4d1453/F2.jpg", "media/5aaecd8df9286f4d1453/F3.jpg", "media/5aaecd8df9286f4d1453/F4.jpg", "media/5aaecd8df9286f4d1453/F5.jpg", "media/5aaecd8df9286f4d1453/F6.jpg", "media/5aaecd8df9286f4d1453/F7.jpg", "media/5aaecd8df9286f4d1453/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dc3ecb3677a2cc5e41350f8680ef9726a0015b1d4c1542ae77425d3b226a110a"}
{"qa_id": "uvo__16dDTSncELA__camera_view_direction_change__01", "video_id": "uvo__16dDTSncELA", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/5aaecd8df9286f4d1453/F0.jpg", "media/5aaecd8df9286f4d1453/F1.jpg", "media/5aaecd8df9286f4d1453/F2.jpg", "media/5aaecd8df9286f4d1453/F3.jpg", "media/5aaecd8df9286f4d1453/F4.jpg", "media/5aaecd8df9286f4d1453/F5.jpg", "media/5aaecd8df9286f4d1453/F6.jpg", "media/5aaecd8df9286f4d1453/F7.jpg", "media/5aaecd8df9286f4d1453/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "480e55fd8a65ac9933e67a2d8f3138ca5da9e9879630fe1c30a3d92892fcd8ff"}
{"qa_id": "uvo__16dDTSncELA__object_world_motion_direction__02", "video_id": "uvo__16dDTSncELA", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in red shirt walking away) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "down", "up", "left"], "frame_paths": ["media/316cf6d600c81017bf26/F0.jpg", "media/316cf6d600c81017bf26/F1.jpg", "media/316cf6d600c81017bf26/F2.jpg", "media/316cf6d600c81017bf26/F3.jpg", "media/316cf6d600c81017bf26/F4.jpg", "media/316cf6d600c81017bf26/F5.jpg", "media/316cf6d600c81017bf26/F6.jpg", "media/316cf6d600c81017bf26/F7.jpg", "media/316cf6d600c81017bf26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in red shirt walking away) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "723052491d2d73016d6cd4c5db5fd52303dd1b11f82d02595708fd51c253a972"}
{"qa_id": "uvo__16dDTSncELA__object_world_motion_direction__03", "video_id": "uvo__16dDTSncELA", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object B (person in blue shirt sitting in a wheelchair) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "right", "left", "up", "forward"], "frame_paths": ["media/105d2f7e30973b594dc8/F0.jpg", "media/105d2f7e30973b594dc8/F1.jpg", "media/105d2f7e30973b594dc8/F2.jpg", "media/105d2f7e30973b594dc8/F3.jpg", "media/105d2f7e30973b594dc8/F4.jpg", "media/105d2f7e30973b594dc8/F5.jpg", "media/105d2f7e30973b594dc8/F6.jpg", "media/105d2f7e30973b594dc8/F7.jpg", "media/105d2f7e30973b594dc8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object B (person in blue shirt sitting in a wheelchair) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. right\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "554629f96d573eab72f2bdf87a1ebc60522295f093f9e3a015b7a564a5f2a361"}
{"qa_id": "uvo__16dDTSncELA__object_travel_distance_comparison__04", "video_id": "uvo__16dDTSncELA", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt walking away) or B (person in blue shirt sitting in a wheelchair)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/a71bb6ebb98d17de6755/F0.jpg", "media/a71bb6ebb98d17de6755/F1.jpg", "media/a71bb6ebb98d17de6755/F2.jpg", "media/a71bb6ebb98d17de6755/F3.jpg", "media/a71bb6ebb98d17de6755/F4.jpg", "media/a71bb6ebb98d17de6755/F5.jpg", "media/a71bb6ebb98d17de6755/F6.jpg", "media/a71bb6ebb98d17de6755/F7.jpg", "media/a71bb6ebb98d17de6755/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt walking away) or B (person in blue shirt sitting in a wheelchair)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eda90cc3e1160335ee602973180e93787623b671e028e77c6baf562369d7f145"}
{"qa_id": "uvo__17iNFBWzqOg__camera_view_direction_change__00", "video_id": "uvo__17iNFBWzqOg", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/791dd0581e64ab1691c0/F0.jpg", "media/791dd0581e64ab1691c0/F1.jpg", "media/791dd0581e64ab1691c0/F2.jpg", "media/791dd0581e64ab1691c0/F3.jpg", "media/791dd0581e64ab1691c0/F4.jpg", "media/791dd0581e64ab1691c0/F5.jpg", "media/791dd0581e64ab1691c0/F6.jpg", "media/791dd0581e64ab1691c0/F7.jpg", "media/791dd0581e64ab1691c0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e64d0963f8222ccd33fac01d2c32f2a7d036cdef24c456b6eaa9e433b71e48fa"}
{"qa_id": "uvo__17v7761cTMU__camera_translation_direction__00", "video_id": "uvo__17v7761cTMU", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "forward", "down", "up"], "frame_paths": ["media/6a10fd00f5879c61a9e8/F0.jpg", "media/6a10fd00f5879c61a9e8/F1.jpg", "media/6a10fd00f5879c61a9e8/F2.jpg", "media/6a10fd00f5879c61a9e8/F3.jpg", "media/6a10fd00f5879c61a9e8/F4.jpg", "media/6a10fd00f5879c61a9e8/F5.jpg", "media/6a10fd00f5879c61a9e8/F6.jpg", "media/6a10fd00f5879c61a9e8/F7.jpg", "media/6a10fd00f5879c61a9e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0224eabf1edb133cf90e6cc68d33bdac02ab13a18e5904a1e48492b4c550b44b"}
{"qa_id": "uvo__17v7761cTMU__camera_object_distance_change__01", "video_id": "uvo__17v7761cTMU", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object B (person wearing a blue t-shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/048364de717d8f1b8018/F0.jpg", "media/048364de717d8f1b8018/F1.jpg", "media/048364de717d8f1b8018/F2.jpg", "media/048364de717d8f1b8018/F3.jpg", "media/048364de717d8f1b8018/F4.jpg", "media/048364de717d8f1b8018/F5.jpg", "media/048364de717d8f1b8018/F6.jpg", "media/048364de717d8f1b8018/F7.jpg", "media/048364de717d8f1b8018/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object B (person wearing a blue t-shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e6bd33ac75bf15dea551a1b7c78253e067906366983aa55c6b1418571836e07"}
{"qa_id": "uvo__17v7761cTMU__object_travel_distance_comparison__02", "video_id": "uvo__17v7761cTMU", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a red long-sleeved shirt and dark pants) or B (person wearing a blue t-shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/d67a4b4c588554c091ff/F0.jpg", "media/d67a4b4c588554c091ff/F1.jpg", "media/d67a4b4c588554c091ff/F2.jpg", "media/d67a4b4c588554c091ff/F3.jpg", "media/d67a4b4c588554c091ff/F4.jpg", "media/d67a4b4c588554c091ff/F5.jpg", "media/d67a4b4c588554c091ff/F6.jpg", "media/d67a4b4c588554c091ff/F7.jpg", "media/d67a4b4c588554c091ff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a red long-sleeved shirt and dark pants) or B (person wearing a blue t-shirt and dark pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "738c911fe7b19a28699df689617e865432ab3f6ef85848b93e10bfa8e862fcb9"}
{"qa_id": "uvo__181Y8iFARNQ__object_pair_distance_change__00", "video_id": "uvo__181Y8iFARNQ", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F6 to F8, did the 3D distance between object A (dancer in red skirt and white top on the right) and object B (dancer in red skirt and white top on the left) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/1866815d441cab6f60c7/F0.jpg", "media/1866815d441cab6f60c7/F1.jpg", "media/1866815d441cab6f60c7/F2.jpg", "media/1866815d441cab6f60c7/F3.jpg", "media/1866815d441cab6f60c7/F4.jpg", "media/1866815d441cab6f60c7/F5.jpg", "media/1866815d441cab6f60c7/F6.jpg", "media/1866815d441cab6f60c7/F7.jpg", "media/1866815d441cab6f60c7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": ["g2", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (dancer in red skirt and white top on the right) and object B (dancer in red skirt and white top on the left) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7ea9851b01b6af10e0a360f7b8279860f9f680ac840288ca5c6b7f3dc18fad47"}
{"qa_id": "uvo__19PvF8GoPp4__camera_translation_direction__00", "video_id": "uvo__19PvF8GoPp4", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "up", "down"], "frame_paths": ["media/f172806b59c1f0a8ff1a/F0.jpg", "media/f172806b59c1f0a8ff1a/F1.jpg", "media/f172806b59c1f0a8ff1a/F2.jpg", "media/f172806b59c1f0a8ff1a/F3.jpg", "media/f172806b59c1f0a8ff1a/F4.jpg", "media/f172806b59c1f0a8ff1a/F5.jpg", "media/f172806b59c1f0a8ff1a/F6.jpg", "media/f172806b59c1f0a8ff1a/F7.jpg", "media/f172806b59c1f0a8ff1a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "42d495577af686fd52ad42be790aca5b4c65d13eeff05998ce2780774a860623"}
{"qa_id": "uvo__19nwtcDcTzA__object_world_motion_direction__00", "video_id": "uvo__19nwtcDcTzA", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (person in dark shirt and red shorts in the center) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "up", "down", "left"], "frame_paths": ["media/2fad473cf71ddb93e865/F0.jpg", "media/2fad473cf71ddb93e865/F1.jpg", "media/2fad473cf71ddb93e865/F2.jpg", "media/2fad473cf71ddb93e865/F3.jpg", "media/2fad473cf71ddb93e865/F4.jpg", "media/2fad473cf71ddb93e865/F5.jpg", "media/2fad473cf71ddb93e865/F6.jpg", "media/2fad473cf71ddb93e865/F7.jpg", "media/2fad473cf71ddb93e865/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (person in dark shirt and red shorts in the center) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e2d537265c908b8ceb1f462d1cf319e190ce5d5e166ed4d2f01b1eb5b1fce85"}
{"qa_id": "uvo__19nwtcDcTzA__camera_object_distance_change__01", "video_id": "uvo__19nwtcDcTzA", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F8, did object B (person in dark shirt and red shorts in the center) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/2fad473cf71ddb93e865/F0.jpg", "media/2fad473cf71ddb93e865/F1.jpg", "media/2fad473cf71ddb93e865/F2.jpg", "media/2fad473cf71ddb93e865/F3.jpg", "media/2fad473cf71ddb93e865/F4.jpg", "media/2fad473cf71ddb93e865/F5.jpg", "media/2fad473cf71ddb93e865/F6.jpg", "media/2fad473cf71ddb93e865/F7.jpg", "media/2fad473cf71ddb93e865/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (person in dark shirt and red shorts in the center) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e5a98038942b1d0cac6f18d9b0913cbc74e3a4c775656bc0958212ab355f0bd"}
{"qa_id": "uvo__19nwtcDcTzA__object_pair_distance_change__02", "video_id": "uvo__19nwtcDcTzA", "task": "object_pair_distance_change", "group": "LLL", "group4": "LLL", "question": "From F5 to F8, did the 3D distance between object A (person in white shirt and dark shorts on the left side) and object B (person in dark shirt and red shorts in the center) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/f83c2a0c387bdb3b37c3/F0.jpg", "media/f83c2a0c387bdb3b37c3/F1.jpg", "media/f83c2a0c387bdb3b37c3/F2.jpg", "media/f83c2a0c387bdb3b37c3/F3.jpg", "media/f83c2a0c387bdb3b37c3/F4.jpg", "media/f83c2a0c387bdb3b37c3/F5.jpg", "media/f83c2a0c387bdb3b37c3/F6.jpg", "media/f83c2a0c387bdb3b37c3/F7.jpg", "media/f83c2a0c387bdb3b37c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person in white shirt and dark shorts on the left side) and object B (person in dark shirt and red shorts in the center) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64c98129608f53b607777c9409ab892316238ee23e8ae61be5584dfa13a750eb"}
{"qa_id": "uvo__19nwtcDcTzA__object_travel_distance_comparison__03", "video_id": "uvo__19nwtcDcTzA", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and dark shorts on the left side) or B (person in dark shirt and red shorts in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/f83c2a0c387bdb3b37c3/F0.jpg", "media/f83c2a0c387bdb3b37c3/F1.jpg", "media/f83c2a0c387bdb3b37c3/F2.jpg", "media/f83c2a0c387bdb3b37c3/F3.jpg", "media/f83c2a0c387bdb3b37c3/F4.jpg", "media/f83c2a0c387bdb3b37c3/F5.jpg", "media/f83c2a0c387bdb3b37c3/F6.jpg", "media/f83c2a0c387bdb3b37c3/F7.jpg", "media/f83c2a0c387bdb3b37c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and dark shorts on the left side) or B (person in dark shirt and red shorts in the center)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e5f6ec9717089c736295a26b82c3f29b2e8df445ae9402df54bace2b30fbffeb"}
{"qa_id": "uvo__1ALF46LXwUU__object_world_motion_direction__00", "video_id": "uvo__1ALF46LXwUU", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in black shirt and white shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "right", "forward", "left"], "frame_paths": ["media/ff605e67d79180b9e1bb/F0.jpg", "media/ff605e67d79180b9e1bb/F1.jpg", "media/ff605e67d79180b9e1bb/F2.jpg", "media/ff605e67d79180b9e1bb/F3.jpg", "media/ff605e67d79180b9e1bb/F4.jpg", "media/ff605e67d79180b9e1bb/F5.jpg", "media/ff605e67d79180b9e1bb/F6.jpg", "media/ff605e67d79180b9e1bb/F7.jpg", "media/ff605e67d79180b9e1bb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in black shirt and white shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a7076e1544041a1763922108b5140161d1753c0be3c2220a26a99f965efe686b"}
{"qa_id": "uvo__1ALF46LXwUU__camera_object_distance_change__01", "video_id": "uvo__1ALF46LXwUU", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (person in black shirt and white shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ff605e67d79180b9e1bb/F0.jpg", "media/ff605e67d79180b9e1bb/F1.jpg", "media/ff605e67d79180b9e1bb/F2.jpg", "media/ff605e67d79180b9e1bb/F3.jpg", "media/ff605e67d79180b9e1bb/F4.jpg", "media/ff605e67d79180b9e1bb/F5.jpg", "media/ff605e67d79180b9e1bb/F6.jpg", "media/ff605e67d79180b9e1bb/F7.jpg", "media/ff605e67d79180b9e1bb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (person in black shirt and white shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5de8cf1a7aadea91668a102b632754c0b118610846c32a2a463ff463e57cdf53"}
{"qa_id": "uvo__1AXnss7WdJI__camera_view_direction_change__00", "video_id": "uvo__1AXnss7WdJI", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/f674dbd84cc61d9fd6ae/F0.jpg", "media/f674dbd84cc61d9fd6ae/F1.jpg", "media/f674dbd84cc61d9fd6ae/F2.jpg", "media/f674dbd84cc61d9fd6ae/F3.jpg", "media/f674dbd84cc61d9fd6ae/F4.jpg", "media/f674dbd84cc61d9fd6ae/F5.jpg", "media/f674dbd84cc61d9fd6ae/F6.jpg", "media/f674dbd84cc61d9fd6ae/F7.jpg", "media/f674dbd84cc61d9fd6ae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "08e062e30d30bd5dfc750194c8aa43281538706fda09276339725fa1f33c3f94"}
{"qa_id": "uvo__1C4SkwtFef0__object_world_motion_direction__00", "video_id": "uvo__1C4SkwtFef0", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (man in blue sleeveless shirt walking on the sidewalk) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "forward", "right", "down"], "frame_paths": ["media/c945a222887205f834b3/F0.jpg", "media/c945a222887205f834b3/F1.jpg", "media/c945a222887205f834b3/F2.jpg", "media/c945a222887205f834b3/F3.jpg", "media/c945a222887205f834b3/F4.jpg", "media/c945a222887205f834b3/F5.jpg", "media/c945a222887205f834b3/F6.jpg", "media/c945a222887205f834b3/F7.jpg", "media/c945a222887205f834b3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (man in blue sleeveless shirt walking on the sidewalk) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f6a6916afa38cf66b596b76256af5a4c8760bf0c34c43ac8f2c81957d0c4615"}
{"qa_id": "uvo__1C4SkwtFef0__camera_object_distance_change__01", "video_id": "uvo__1C4SkwtFef0", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did object B (man in blue sleeveless shirt walking on the sidewalk) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c945a222887205f834b3/F0.jpg", "media/c945a222887205f834b3/F1.jpg", "media/c945a222887205f834b3/F2.jpg", "media/c945a222887205f834b3/F3.jpg", "media/c945a222887205f834b3/F4.jpg", "media/c945a222887205f834b3/F5.jpg", "media/c945a222887205f834b3/F6.jpg", "media/c945a222887205f834b3/F7.jpg", "media/c945a222887205f834b3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object B (man in blue sleeveless shirt walking on the sidewalk) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34c769e971e4dd46bae01675d2848da33dce3df5861577e4edd661b2960016e6"}
{"qa_id": "uvo__1C4SkwtFef0__object_pair_distance_change__02", "video_id": "uvo__1C4SkwtFef0", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, did the 3D distance between object A (man in black shirt and cap standing near a musician) and object B (man in blue sleeveless shirt walking on the sidewalk) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/a02c973abcbfaf18c1cb/F0.jpg", "media/a02c973abcbfaf18c1cb/F1.jpg", "media/a02c973abcbfaf18c1cb/F2.jpg", "media/a02c973abcbfaf18c1cb/F3.jpg", "media/a02c973abcbfaf18c1cb/F4.jpg", "media/a02c973abcbfaf18c1cb/F5.jpg", "media/a02c973abcbfaf18c1cb/F6.jpg", "media/a02c973abcbfaf18c1cb/F7.jpg", "media/a02c973abcbfaf18c1cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did the 3D distance between object A (man in black shirt and cap standing near a musician) and object B (man in blue sleeveless shirt walking on the sidewalk) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dd3977135d14a20899f982e89cfe31c13e2603038d9a0c75b0fb3514de4791ba"}
{"qa_id": "uvo__1C4SkwtFef0__object_travel_distance_comparison__03", "video_id": "uvo__1C4SkwtFef0", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in black shirt and cap standing near a musician) or B (man in blue sleeveless shirt walking on the sidewalk)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/a02c973abcbfaf18c1cb/F0.jpg", "media/a02c973abcbfaf18c1cb/F1.jpg", "media/a02c973abcbfaf18c1cb/F2.jpg", "media/a02c973abcbfaf18c1cb/F3.jpg", "media/a02c973abcbfaf18c1cb/F4.jpg", "media/a02c973abcbfaf18c1cb/F5.jpg", "media/a02c973abcbfaf18c1cb/F6.jpg", "media/a02c973abcbfaf18c1cb/F7.jpg", "media/a02c973abcbfaf18c1cb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in black shirt and cap standing near a musician) or B (man in blue sleeveless shirt walking on the sidewalk)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2b3f20c805bac66ed3958481a87d45f3587297292ab7998b55c808a8e7fcac7"}
{"qa_id": "uvo__1C8rdJEBgbU__camera_view_direction_change__00", "video_id": "uvo__1C8rdJEBgbU", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/864538d9be5993b7aae0/F0.jpg", "media/864538d9be5993b7aae0/F1.jpg", "media/864538d9be5993b7aae0/F2.jpg", "media/864538d9be5993b7aae0/F3.jpg", "media/864538d9be5993b7aae0/F4.jpg", "media/864538d9be5993b7aae0/F5.jpg", "media/864538d9be5993b7aae0/F6.jpg", "media/864538d9be5993b7aae0/F7.jpg", "media/864538d9be5993b7aae0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7cb4feef62c3d9efd6762a749f2132199fae31e2272a2ff8d2b33844b5eac135"}
{"qa_id": "uvo__1CTlE-tPD_A__camera_translation_direction__00", "video_id": "uvo__1CTlE-tPD_A", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "up", "backward", "right", "down"], "frame_paths": ["media/0ca54c940abca2644542/F0.jpg", "media/0ca54c940abca2644542/F1.jpg", "media/0ca54c940abca2644542/F2.jpg", "media/0ca54c940abca2644542/F3.jpg", "media/0ca54c940abca2644542/F4.jpg", "media/0ca54c940abca2644542/F5.jpg", "media/0ca54c940abca2644542/F6.jpg", "media/0ca54c940abca2644542/F7.jpg", "media/0ca54c940abca2644542/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. up\nD. backward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9a3290e1fe7c29d8403e786088f5925c8f54f1cf8be76e89f3f33a84b9fd820"}
{"qa_id": "uvo__1CTlE-tPD_A__camera_view_direction_change__01", "video_id": "uvo__1CTlE-tPD_A", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/0ca54c940abca2644542/F0.jpg", "media/0ca54c940abca2644542/F1.jpg", "media/0ca54c940abca2644542/F2.jpg", "media/0ca54c940abca2644542/F3.jpg", "media/0ca54c940abca2644542/F4.jpg", "media/0ca54c940abca2644542/F5.jpg", "media/0ca54c940abca2644542/F6.jpg", "media/0ca54c940abca2644542/F7.jpg", "media/0ca54c940abca2644542/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "396ce893acdee73d9f632ef9b2a536c9073a438617f550aef2dc7a7ed2a1a220"}
{"qa_id": "uvo__1DLRj_ytlMQ__camera_object_distance_change__00", "video_id": "uvo__1DLRj_ytlMQ", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (person in black shirt and blue shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f58eb8d2d519a3d5b84b/F0.jpg", "media/f58eb8d2d519a3d5b84b/F1.jpg", "media/f58eb8d2d519a3d5b84b/F2.jpg", "media/f58eb8d2d519a3d5b84b/F3.jpg", "media/f58eb8d2d519a3d5b84b/F4.jpg", "media/f58eb8d2d519a3d5b84b/F5.jpg", "media/f58eb8d2d519a3d5b84b/F6.jpg", "media/f58eb8d2d519a3d5b84b/F7.jpg", "media/f58eb8d2d519a3d5b84b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in black shirt and blue shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aae658051a059ff2ddbab301dc13c35c87ad891c2ae06056d3e9862a4c487624"}
{"qa_id": "uvo__1DLRj_ytlMQ__object_pair_distance_change__01", "video_id": "uvo__1DLRj_ytlMQ", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did the 3D distance between object A (person in black shirt and blue shorts) and object B (person in black jersey with number 32) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d0e465920be7aedbefd0/F0.jpg", "media/d0e465920be7aedbefd0/F1.jpg", "media/d0e465920be7aedbefd0/F2.jpg", "media/d0e465920be7aedbefd0/F3.jpg", "media/d0e465920be7aedbefd0/F4.jpg", "media/d0e465920be7aedbefd0/F5.jpg", "media/d0e465920be7aedbefd0/F6.jpg", "media/d0e465920be7aedbefd0/F7.jpg", "media/d0e465920be7aedbefd0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did the 3D distance between object A (person in black shirt and blue shorts) and object B (person in black jersey with number 32) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2d8b23a07988e45b59c726e9390d51950b472468df774b24002c3dfc8a01a5ad"}
{"qa_id": "uvo__1DLRj_ytlMQ__object_travel_distance_comparison__02", "video_id": "uvo__1DLRj_ytlMQ", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black shirt and blue shorts) or B (person in black jersey with number 32)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/d0e465920be7aedbefd0/F0.jpg", "media/d0e465920be7aedbefd0/F1.jpg", "media/d0e465920be7aedbefd0/F2.jpg", "media/d0e465920be7aedbefd0/F3.jpg", "media/d0e465920be7aedbefd0/F4.jpg", "media/d0e465920be7aedbefd0/F5.jpg", "media/d0e465920be7aedbefd0/F6.jpg", "media/d0e465920be7aedbefd0/F7.jpg", "media/d0e465920be7aedbefd0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black shirt and blue shorts) or B (person in black jersey with number 32)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aef52a40332278c6c086a031e9c7e7ba639e38221ffc344e60b227ef6bd8e396"}
{"qa_id": "uvo__1EEIjxNmFgk__camera_view_direction_change__00", "video_id": "uvo__1EEIjxNmFgk", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/21ffe9a289bcebaed8c3/F0.jpg", "media/21ffe9a289bcebaed8c3/F1.jpg", "media/21ffe9a289bcebaed8c3/F2.jpg", "media/21ffe9a289bcebaed8c3/F3.jpg", "media/21ffe9a289bcebaed8c3/F4.jpg", "media/21ffe9a289bcebaed8c3/F5.jpg", "media/21ffe9a289bcebaed8c3/F6.jpg", "media/21ffe9a289bcebaed8c3/F7.jpg", "media/21ffe9a289bcebaed8c3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "acd59f5999362d81d16fc7a769204a3775b0696b58f897bc7e67bc16ca07099b"}
{"qa_id": "uvo__1EEIjxNmFgk__object_pair_distance_change__01", "video_id": "uvo__1EEIjxNmFgk", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (hockey player in black jersey and orange pants) and object B (hockey player in green jersey) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/b192d5b8833c4e1dcb56/F0.jpg", "media/b192d5b8833c4e1dcb56/F1.jpg", "media/b192d5b8833c4e1dcb56/F2.jpg", "media/b192d5b8833c4e1dcb56/F3.jpg", "media/b192d5b8833c4e1dcb56/F4.jpg", "media/b192d5b8833c4e1dcb56/F5.jpg", "media/b192d5b8833c4e1dcb56/F6.jpg", "media/b192d5b8833c4e1dcb56/F7.jpg", "media/b192d5b8833c4e1dcb56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (hockey player in black jersey and orange pants) and object B (hockey player in green jersey) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "da2d4c2abfcef66cd9d4a9af24d08ae25620ee7d33d07d4f3ba57a1eecd79ca3"}
{"qa_id": "uvo__1EEIjxNmFgk__object_travel_distance_comparison__02", "video_id": "uvo__1EEIjxNmFgk", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (hockey player in black jersey and orange pants) or B (hockey player in green jersey)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/b192d5b8833c4e1dcb56/F0.jpg", "media/b192d5b8833c4e1dcb56/F1.jpg", "media/b192d5b8833c4e1dcb56/F2.jpg", "media/b192d5b8833c4e1dcb56/F3.jpg", "media/b192d5b8833c4e1dcb56/F4.jpg", "media/b192d5b8833c4e1dcb56/F5.jpg", "media/b192d5b8833c4e1dcb56/F6.jpg", "media/b192d5b8833c4e1dcb56/F7.jpg", "media/b192d5b8833c4e1dcb56/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (hockey player in black jersey and orange pants) or B (hockey player in green jersey)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36965d2b362138213d5ae3edfcee972afe0b44a5c28c1176b030ec6baafbd1dd"}
{"qa_id": "uvo__1EVYUcLNXbE__object_world_motion_direction__00", "video_id": "uvo__1EVYUcLNXbE", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object B (person in green shirt and shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "down", "backward", "left"], "frame_paths": ["media/066d10630e6ed91c35a7/F0.jpg", "media/066d10630e6ed91c35a7/F1.jpg", "media/066d10630e6ed91c35a7/F2.jpg", "media/066d10630e6ed91c35a7/F3.jpg", "media/066d10630e6ed91c35a7/F4.jpg", "media/066d10630e6ed91c35a7/F5.jpg", "media/066d10630e6ed91c35a7/F6.jpg", "media/066d10630e6ed91c35a7/F7.jpg", "media/066d10630e6ed91c35a7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object B (person in green shirt and shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. down\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f8291da5fbd0871980bf867cdd2131e08c2a83ceece2f20a2c23b1dde2db403b"}
{"qa_id": "uvo__1EVYUcLNXbE__object_pair_distance_change__01", "video_id": "uvo__1EVYUcLNXbE", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did the 3D distance between object A (black dog with a white bowl) and object B (person in green shirt and shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/800d7ba069bc163707d0/F0.jpg", "media/800d7ba069bc163707d0/F1.jpg", "media/800d7ba069bc163707d0/F2.jpg", "media/800d7ba069bc163707d0/F3.jpg", "media/800d7ba069bc163707d0/F4.jpg", "media/800d7ba069bc163707d0/F5.jpg", "media/800d7ba069bc163707d0/F6.jpg", "media/800d7ba069bc163707d0/F7.jpg", "media/800d7ba069bc163707d0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (black dog with a white bowl) and object B (person in green shirt and shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bfb306fb7365726148dcf9f6d3cbc616dd6bbb932b5f14306a4dd7d0f94aceb7"}
{"qa_id": "uvo__1EVYUcLNXbE__object_travel_distance_comparison__02", "video_id": "uvo__1EVYUcLNXbE", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black dog with a white bowl) or B (person in green shirt and shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/800d7ba069bc163707d0/F0.jpg", "media/800d7ba069bc163707d0/F1.jpg", "media/800d7ba069bc163707d0/F2.jpg", "media/800d7ba069bc163707d0/F3.jpg", "media/800d7ba069bc163707d0/F4.jpg", "media/800d7ba069bc163707d0/F5.jpg", "media/800d7ba069bc163707d0/F6.jpg", "media/800d7ba069bc163707d0/F7.jpg", "media/800d7ba069bc163707d0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (black dog with a white bowl) or B (person in green shirt and shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c58fd7ed969f877c952ba885c21e49c0879311fcb5ed38d8aff75a75432a75f2"}
{"qa_id": "uvo__1EdD1OkUrLs__camera_view_direction_change__00", "video_id": "uvo__1EdD1OkUrLs", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/84d2f4d3007d0058a805/F0.jpg", "media/84d2f4d3007d0058a805/F1.jpg", "media/84d2f4d3007d0058a805/F2.jpg", "media/84d2f4d3007d0058a805/F3.jpg", "media/84d2f4d3007d0058a805/F4.jpg", "media/84d2f4d3007d0058a805/F5.jpg", "media/84d2f4d3007d0058a805/F6.jpg", "media/84d2f4d3007d0058a805/F7.jpg", "media/84d2f4d3007d0058a805/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa16a7b0e3ca16496f3925cd225c2680848db8fe36c17867e87269224a69a1ff"}
{"qa_id": "uvo__1EdD1OkUrLs__object_pair_distance_change__01", "video_id": "uvo__1EdD1OkUrLs", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (man in green jersey and khaki pants) and object B (boy in black hoodie seen from behind) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/b7587f70b90c76ae530b/F0.jpg", "media/b7587f70b90c76ae530b/F1.jpg", "media/b7587f70b90c76ae530b/F2.jpg", "media/b7587f70b90c76ae530b/F3.jpg", "media/b7587f70b90c76ae530b/F4.jpg", "media/b7587f70b90c76ae530b/F5.jpg", "media/b7587f70b90c76ae530b/F6.jpg", "media/b7587f70b90c76ae530b/F7.jpg", "media/b7587f70b90c76ae530b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (man in green jersey and khaki pants) and object B (boy in black hoodie seen from behind) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8985cd7731746d60e1d274577b5137548c467db5cb7573b66698815248e4542a"}
{"qa_id": "uvo__1Ep1RNI0b0U__camera_translation_direction__00", "video_id": "uvo__1Ep1RNI0b0U", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "backward", "down", "left"], "frame_paths": ["media/a78605ee56842397b0de/F0.jpg", "media/a78605ee56842397b0de/F1.jpg", "media/a78605ee56842397b0de/F2.jpg", "media/a78605ee56842397b0de/F3.jpg", "media/a78605ee56842397b0de/F4.jpg", "media/a78605ee56842397b0de/F5.jpg", "media/a78605ee56842397b0de/F6.jpg", "media/a78605ee56842397b0de/F7.jpg", "media/a78605ee56842397b0de/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fdc304ac83f011c013321abe11e01af99ec104ea7fa82f4862f8e0a1704fa2e5"}
{"qa_id": "uvo__1Ep1RNI0b0U__camera_object_distance_change__01", "video_id": "uvo__1Ep1RNI0b0U", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, did object A (orange and teal parachute canopy) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/afdaedb51b5a0b618a48/F0.jpg", "media/afdaedb51b5a0b618a48/F1.jpg", "media/afdaedb51b5a0b618a48/F2.jpg", "media/afdaedb51b5a0b618a48/F3.jpg", "media/afdaedb51b5a0b618a48/F4.jpg", "media/afdaedb51b5a0b618a48/F5.jpg", "media/afdaedb51b5a0b618a48/F6.jpg", "media/afdaedb51b5a0b618a48/F7.jpg", "media/afdaedb51b5a0b618a48/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (orange and teal parachute canopy) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ace24e0c7ec778fd7d7c14294881aaadc4e855958a291ce397dd658323526bce"}
{"qa_id": "uvo__1Evf8RZdLOo__camera_translation_direction__00", "video_id": "uvo__1Evf8RZdLOo", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "down", "up", "left"], "frame_paths": ["media/3b8a607389850352a3aa/F0.jpg", "media/3b8a607389850352a3aa/F1.jpg", "media/3b8a607389850352a3aa/F2.jpg", "media/3b8a607389850352a3aa/F3.jpg", "media/3b8a607389850352a3aa/F4.jpg", "media/3b8a607389850352a3aa/F5.jpg", "media/3b8a607389850352a3aa/F6.jpg", "media/3b8a607389850352a3aa/F7.jpg", "media/3b8a607389850352a3aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "002a4146efa7231bd10fad58b3831c7cbae2ec6ac06a88f57e65b30e13b05104"}
{"qa_id": "uvo__1Evf8RZdLOo__camera_view_direction_change__01", "video_id": "uvo__1Evf8RZdLOo", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/3b8a607389850352a3aa/F0.jpg", "media/3b8a607389850352a3aa/F1.jpg", "media/3b8a607389850352a3aa/F2.jpg", "media/3b8a607389850352a3aa/F3.jpg", "media/3b8a607389850352a3aa/F4.jpg", "media/3b8a607389850352a3aa/F5.jpg", "media/3b8a607389850352a3aa/F6.jpg", "media/3b8a607389850352a3aa/F7.jpg", "media/3b8a607389850352a3aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1dbe99d954dda87041f1ab0a8a6bf827534985b550ea6cb13ae4e1220d0a1cad"}
{"qa_id": "uvo__1F8DhAMABpU__camera_view_direction_change__00", "video_id": "uvo__1F8DhAMABpU", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/2c2049275be92ca88fae/F0.jpg", "media/2c2049275be92ca88fae/F1.jpg", "media/2c2049275be92ca88fae/F2.jpg", "media/2c2049275be92ca88fae/F3.jpg", "media/2c2049275be92ca88fae/F4.jpg", "media/2c2049275be92ca88fae/F5.jpg", "media/2c2049275be92ca88fae/F6.jpg", "media/2c2049275be92ca88fae/F7.jpg", "media/2c2049275be92ca88fae/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b05c59c5ff5c9031ea68746d67c85c07217775e2ecc9e157cbe29c5a6487d912"}
{"qa_id": "uvo__1FaC6nzR2eY__object_world_motion_direction__00", "video_id": "uvo__1FaC6nzR2eY", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (black dog running on a snowy path) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "left", "backward", "forward"], "frame_paths": ["media/1a874684d3cabf865946/F0.jpg", "media/1a874684d3cabf865946/F1.jpg", "media/1a874684d3cabf865946/F2.jpg", "media/1a874684d3cabf865946/F3.jpg", "media/1a874684d3cabf865946/F4.jpg", "media/1a874684d3cabf865946/F5.jpg", "media/1a874684d3cabf865946/F6.jpg", "media/1a874684d3cabf865946/F7.jpg", "media/1a874684d3cabf865946/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (black dog running on a snowy path) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "992fbdb28c6f2fcf2e18fefa9df285fa8a19482361c8d6db2b309552846430b5"}
{"qa_id": "uvo__1FaC6nzR2eY__camera_object_distance_change__01", "video_id": "uvo__1FaC6nzR2eY", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did object A (black dog running on a snowy path) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/1a874684d3cabf865946/F0.jpg", "media/1a874684d3cabf865946/F1.jpg", "media/1a874684d3cabf865946/F2.jpg", "media/1a874684d3cabf865946/F3.jpg", "media/1a874684d3cabf865946/F4.jpg", "media/1a874684d3cabf865946/F5.jpg", "media/1a874684d3cabf865946/F6.jpg", "media/1a874684d3cabf865946/F7.jpg", "media/1a874684d3cabf865946/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (black dog running on a snowy path) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e3e5ba315c39aedaee2618a7c05580b32bd9429e5a904a6c1d1b88caa0820574"}
{"qa_id": "uvo__1Gez2PibPlg__object_world_motion_direction__00", "video_id": "uvo__1Gez2PibPlg", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (small orange boat moving on the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "backward", "forward", "left"], "frame_paths": ["media/a1d7db8c144c396b41b5/F0.jpg", "media/a1d7db8c144c396b41b5/F1.jpg", "media/a1d7db8c144c396b41b5/F2.jpg", "media/a1d7db8c144c396b41b5/F3.jpg", "media/a1d7db8c144c396b41b5/F4.jpg", "media/a1d7db8c144c396b41b5/F5.jpg", "media/a1d7db8c144c396b41b5/F6.jpg", "media/a1d7db8c144c396b41b5/F7.jpg", "media/a1d7db8c144c396b41b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (small orange boat moving on the water) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74659d8de4668aa973ab2c81dc072bb7a80da4d7d7bd3d624f293fd9e51631b1"}
{"qa_id": "uvo__1Gez2PibPlg__camera_object_distance_change__01", "video_id": "uvo__1Gez2PibPlg", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (small orange boat moving on the water) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a1d7db8c144c396b41b5/F0.jpg", "media/a1d7db8c144c396b41b5/F1.jpg", "media/a1d7db8c144c396b41b5/F2.jpg", "media/a1d7db8c144c396b41b5/F3.jpg", "media/a1d7db8c144c396b41b5/F4.jpg", "media/a1d7db8c144c396b41b5/F5.jpg", "media/a1d7db8c144c396b41b5/F6.jpg", "media/a1d7db8c144c396b41b5/F7.jpg", "media/a1d7db8c144c396b41b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (small orange boat moving on the water) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c836e621345a81ed95c41ecf55aa3323923a08fe7841448b2436a47f98abc714"}
{"qa_id": "uvo__1Hi1TB8Pjfs__camera_view_direction_change__00", "video_id": "uvo__1Hi1TB8Pjfs", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/b6a334739d78630ab807/F0.jpg", "media/b6a334739d78630ab807/F1.jpg", "media/b6a334739d78630ab807/F2.jpg", "media/b6a334739d78630ab807/F3.jpg", "media/b6a334739d78630ab807/F4.jpg", "media/b6a334739d78630ab807/F5.jpg", "media/b6a334739d78630ab807/F6.jpg", "media/b6a334739d78630ab807/F7.jpg", "media/b6a334739d78630ab807/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__1HyfE5t-h5o__camera_translation_direction__00", "video_id": "uvo__1HyfE5t-h5o", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "forward", "right", "up", "left"], "frame_paths": ["media/b9ab31c0b6d26737905f/F0.jpg", "media/b9ab31c0b6d26737905f/F1.jpg", "media/b9ab31c0b6d26737905f/F2.jpg", "media/b9ab31c0b6d26737905f/F3.jpg", "media/b9ab31c0b6d26737905f/F4.jpg", "media/b9ab31c0b6d26737905f/F5.jpg", "media/b9ab31c0b6d26737905f/F6.jpg", "media/b9ab31c0b6d26737905f/F7.jpg", "media/b9ab31c0b6d26737905f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. forward\nD. right\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bd3e1639cd4ea9b573e3ca3188921eb66985b021c491e5b10ca06ae82f76c83b"}
{"qa_id": "uvo__1HyfE5t-h5o__object_world_motion_direction__01", "video_id": "uvo__1HyfE5t-h5o", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in orange shirt riding a motorcycle) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "backward", "left", "down"], "frame_paths": ["media/e5dbd7eed04013f34fec/F0.jpg", "media/e5dbd7eed04013f34fec/F1.jpg", "media/e5dbd7eed04013f34fec/F2.jpg", "media/e5dbd7eed04013f34fec/F3.jpg", "media/e5dbd7eed04013f34fec/F4.jpg", "media/e5dbd7eed04013f34fec/F5.jpg", "media/e5dbd7eed04013f34fec/F6.jpg", "media/e5dbd7eed04013f34fec/F7.jpg", "media/e5dbd7eed04013f34fec/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F7, in which dominant direction did object A (person in orange shirt riding a motorcycle) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. backward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b69fc65bc6af4cea4f81a47deba2fbcd93b2e37d974847c13468336e4f7fdbb9"}
{"qa_id": "uvo__1IkQds-3rCY__camera_view_direction_change__00", "video_id": "uvo__1IkQds-3rCY", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/18f69e91608234e364ca/F0.jpg", "media/18f69e91608234e364ca/F1.jpg", "media/18f69e91608234e364ca/F2.jpg", "media/18f69e91608234e364ca/F3.jpg", "media/18f69e91608234e364ca/F4.jpg", "media/18f69e91608234e364ca/F5.jpg", "media/18f69e91608234e364ca/F6.jpg", "media/18f69e91608234e364ca/F7.jpg", "media/18f69e91608234e364ca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f8424c0b08354e78a0af2158ce3016d19c692a05492f95ce15f4393301f02e3"}
{"qa_id": "uvo__1J109MGAspk__camera_view_direction_change__00", "video_id": "uvo__1J109MGAspk", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/4bb0a358fc23e347194e/F0.jpg", "media/4bb0a358fc23e347194e/F1.jpg", "media/4bb0a358fc23e347194e/F2.jpg", "media/4bb0a358fc23e347194e/F3.jpg", "media/4bb0a358fc23e347194e/F4.jpg", "media/4bb0a358fc23e347194e/F5.jpg", "media/4bb0a358fc23e347194e/F6.jpg", "media/4bb0a358fc23e347194e/F7.jpg", "media/4bb0a358fc23e347194e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "13dac79e2013fbcd040093fc210e0a05f869f71c449dfee4a4d2a9a58dd4088f"}
{"qa_id": "uvo__1JK-yDbqK_M__camera_translation_direction__00", "video_id": "uvo__1JK-yDbqK_M", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "right", "forward", "down", "backward"], "frame_paths": ["media/2870d1d679d4e24f3d6f/F0.jpg", "media/2870d1d679d4e24f3d6f/F1.jpg", "media/2870d1d679d4e24f3d6f/F2.jpg", "media/2870d1d679d4e24f3d6f/F3.jpg", "media/2870d1d679d4e24f3d6f/F4.jpg", "media/2870d1d679d4e24f3d6f/F5.jpg", "media/2870d1d679d4e24f3d6f/F6.jpg", "media/2870d1d679d4e24f3d6f/F7.jpg", "media/2870d1d679d4e24f3d6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. right\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa392555c22e014a8c7ec16946ec4f7ee7a13e71279afbb19286ed3f75ff0599"}
{"qa_id": "uvo__1JUL-MDQf3k__camera_view_direction_change__00", "video_id": "uvo__1JUL-MDQf3k", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/efe6a7020180b2e36f5d/F0.jpg", "media/efe6a7020180b2e36f5d/F1.jpg", "media/efe6a7020180b2e36f5d/F2.jpg", "media/efe6a7020180b2e36f5d/F3.jpg", "media/efe6a7020180b2e36f5d/F4.jpg", "media/efe6a7020180b2e36f5d/F5.jpg", "media/efe6a7020180b2e36f5d/F6.jpg", "media/efe6a7020180b2e36f5d/F7.jpg", "media/efe6a7020180b2e36f5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40042d71fd53d20c12062dc9329db03c4e6db8d1df5fac06f2e7bfe3ea843cc3"}
{"qa_id": "uvo__1JhYwcmxvtY__camera_translation_direction__00", "video_id": "uvo__1JhYwcmxvtY", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "right", "up", "backward", "down"], "frame_paths": ["media/755988c78fcf9fe0eefd/F0.jpg", "media/755988c78fcf9fe0eefd/F1.jpg", "media/755988c78fcf9fe0eefd/F2.jpg", "media/755988c78fcf9fe0eefd/F3.jpg", "media/755988c78fcf9fe0eefd/F4.jpg", "media/755988c78fcf9fe0eefd/F5.jpg", "media/755988c78fcf9fe0eefd/F6.jpg", "media/755988c78fcf9fe0eefd/F7.jpg", "media/755988c78fcf9fe0eefd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. right\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eac5b1185cbc4e3f76192a8c643c440a5eee37c04a9328f6ccb0b5ca18da8218"}
{"qa_id": "uvo__1K4FIaiPL6I__camera_translation_direction__00", "video_id": "uvo__1K4FIaiPL6I", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "down", "left", "backward"], "frame_paths": ["media/f32895c9d12a895cd015/F0.jpg", "media/f32895c9d12a895cd015/F1.jpg", "media/f32895c9d12a895cd015/F2.jpg", "media/f32895c9d12a895cd015/F3.jpg", "media/f32895c9d12a895cd015/F4.jpg", "media/f32895c9d12a895cd015/F5.jpg", "media/f32895c9d12a895cd015/F6.jpg", "media/f32895c9d12a895cd015/F7.jpg", "media/f32895c9d12a895cd015/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d0a94f0baa6bf23c66f3a215efa81f5daa5810b94239e1fc151eb6d7dc881cb"}
{"qa_id": "uvo__1K4FIaiPL6I__object_world_motion_direction__01", "video_id": "uvo__1K4FIaiPL6I", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person wearing an orange safety vest and camouflage pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "forward", "left", "down"], "frame_paths": ["media/0ce176bf0d61284ecce2/F0.jpg", "media/0ce176bf0d61284ecce2/F1.jpg", "media/0ce176bf0d61284ecce2/F2.jpg", "media/0ce176bf0d61284ecce2/F3.jpg", "media/0ce176bf0d61284ecce2/F4.jpg", "media/0ce176bf0d61284ecce2/F5.jpg", "media/0ce176bf0d61284ecce2/F6.jpg", "media/0ce176bf0d61284ecce2/F7.jpg", "media/0ce176bf0d61284ecce2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person wearing an orange safety vest and camouflage pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a0f5681999f91bcfef65e911e3b374647bfe911de8b7249ab0b4e94fdba48dd"}
{"qa_id": "uvo__1K5OXfQF4NA__camera_translation_direction__00", "video_id": "uvo__1K5OXfQF4NA", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "down", "up"], "frame_paths": ["media/823f0ecfcbc9279ded5d/F0.jpg", "media/823f0ecfcbc9279ded5d/F1.jpg", "media/823f0ecfcbc9279ded5d/F2.jpg", "media/823f0ecfcbc9279ded5d/F3.jpg", "media/823f0ecfcbc9279ded5d/F4.jpg", "media/823f0ecfcbc9279ded5d/F5.jpg", "media/823f0ecfcbc9279ded5d/F6.jpg", "media/823f0ecfcbc9279ded5d/F7.jpg", "media/823f0ecfcbc9279ded5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "716c3488b044a1b6e6dc675722e863214194a41a3cbaa8f0564ed1ad96c00ffc"}
{"qa_id": "uvo__1K5OXfQF4NA__camera_view_direction_change__01", "video_id": "uvo__1K5OXfQF4NA", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/823f0ecfcbc9279ded5d/F0.jpg", "media/823f0ecfcbc9279ded5d/F1.jpg", "media/823f0ecfcbc9279ded5d/F2.jpg", "media/823f0ecfcbc9279ded5d/F3.jpg", "media/823f0ecfcbc9279ded5d/F4.jpg", "media/823f0ecfcbc9279ded5d/F5.jpg", "media/823f0ecfcbc9279ded5d/F6.jpg", "media/823f0ecfcbc9279ded5d/F7.jpg", "media/823f0ecfcbc9279ded5d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db2dabb9c09d768aef279d407512cee59fa5581206d28df2cf3badd611191af1"}
{"qa_id": "uvo__1KO4KH5i1rk__camera_view_direction_change__00", "video_id": "uvo__1KO4KH5i1rk", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/8782f18e551cd1adbb0b/F0.jpg", "media/8782f18e551cd1adbb0b/F1.jpg", "media/8782f18e551cd1adbb0b/F2.jpg", "media/8782f18e551cd1adbb0b/F3.jpg", "media/8782f18e551cd1adbb0b/F4.jpg", "media/8782f18e551cd1adbb0b/F5.jpg", "media/8782f18e551cd1adbb0b/F6.jpg", "media/8782f18e551cd1adbb0b/F7.jpg", "media/8782f18e551cd1adbb0b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a2efb8d666b97e18b98daa17cd0adf9647759854ba1c72dc0408a9f87df1c43e"}
{"qa_id": "uvo__1LS1eWN_TbE__camera_translation_direction__00", "video_id": "uvo__1LS1eWN_TbE", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "forward", "backward", "left"], "frame_paths": ["media/5aad0425c064db41cd50/F0.jpg", "media/5aad0425c064db41cd50/F1.jpg", "media/5aad0425c064db41cd50/F2.jpg", "media/5aad0425c064db41cd50/F3.jpg", "media/5aad0425c064db41cd50/F4.jpg", "media/5aad0425c064db41cd50/F5.jpg", "media/5aad0425c064db41cd50/F6.jpg", "media/5aad0425c064db41cd50/F7.jpg", "media/5aad0425c064db41cd50/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6c0ef2d49db5f7ec72173ea451a5f999b981839daaf344361eb8032ea0eb6e3f"}
{"qa_id": "uvo__1LS1eWN_TbE__camera_object_distance_change__01", "video_id": "uvo__1LS1eWN_TbE", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, did object A (light blue paraglider canopy) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/5090e63f6ddfd15786a7/F0.jpg", "media/5090e63f6ddfd15786a7/F1.jpg", "media/5090e63f6ddfd15786a7/F2.jpg", "media/5090e63f6ddfd15786a7/F3.jpg", "media/5090e63f6ddfd15786a7/F4.jpg", "media/5090e63f6ddfd15786a7/F5.jpg", "media/5090e63f6ddfd15786a7/F6.jpg", "media/5090e63f6ddfd15786a7/F7.jpg", "media/5090e63f6ddfd15786a7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (light blue paraglider canopy) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ad8657083572d65d7388a4f936e7902673dda86cc0440810f4bb531c2ca1a19c"}
{"qa_id": "uvo__1MGRbeRjh58__camera_view_direction_change__00", "video_id": "uvo__1MGRbeRjh58", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/d28f6f6aa209146632df/F0.jpg", "media/d28f6f6aa209146632df/F1.jpg", "media/d28f6f6aa209146632df/F2.jpg", "media/d28f6f6aa209146632df/F3.jpg", "media/d28f6f6aa209146632df/F4.jpg", "media/d28f6f6aa209146632df/F5.jpg", "media/d28f6f6aa209146632df/F6.jpg", "media/d28f6f6aa209146632df/F7.jpg", "media/d28f6f6aa209146632df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "139d78c609346a36b31a51baa6a9cfbf5b99d316434cbd3d4bd9ef34f83af385"}
{"qa_id": "uvo__1MslzwXZ6o8__camera_translation_direction__00", "video_id": "uvo__1MslzwXZ6o8", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "right", "backward", "down", "forward"], "frame_paths": ["media/5730953880257d711cf6/F0.jpg", "media/5730953880257d711cf6/F1.jpg", "media/5730953880257d711cf6/F2.jpg", "media/5730953880257d711cf6/F3.jpg", "media/5730953880257d711cf6/F4.jpg", "media/5730953880257d711cf6/F5.jpg", "media/5730953880257d711cf6/F6.jpg", "media/5730953880257d711cf6/F7.jpg", "media/5730953880257d711cf6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. right\nD. backward\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5eef8a81f65971fc5dde755f89031ca1472815b8d54c5a51aa45112aa43f326b"}
{"qa_id": "uvo__1N3XzBmArWU__camera_view_direction_change__00", "video_id": "uvo__1N3XzBmArWU", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/d59427c6abd7c4eb568d/F0.jpg", "media/d59427c6abd7c4eb568d/F1.jpg", "media/d59427c6abd7c4eb568d/F2.jpg", "media/d59427c6abd7c4eb568d/F3.jpg", "media/d59427c6abd7c4eb568d/F4.jpg", "media/d59427c6abd7c4eb568d/F5.jpg", "media/d59427c6abd7c4eb568d/F6.jpg", "media/d59427c6abd7c4eb568d/F7.jpg", "media/d59427c6abd7c4eb568d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cd09fefdd522823df42332c8588143a44dff98cf96b1cb99b65bdd6b1d0fed25"}
{"qa_id": "uvo__1PQcIT6UuZo__camera_translation_direction__00", "video_id": "uvo__1PQcIT6UuZo", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "up", "left", "backward"], "frame_paths": ["media/5ed0eb787187470707b5/F0.jpg", "media/5ed0eb787187470707b5/F1.jpg", "media/5ed0eb787187470707b5/F2.jpg", "media/5ed0eb787187470707b5/F3.jpg", "media/5ed0eb787187470707b5/F4.jpg", "media/5ed0eb787187470707b5/F5.jpg", "media/5ed0eb787187470707b5/F6.jpg", "media/5ed0eb787187470707b5/F7.jpg", "media/5ed0eb787187470707b5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. up\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91b5200f253c5ddb8202be77c72125de02a7a3aa1cfd1088495b73d32a4ce3c2"}
{"qa_id": "uvo__1PQcIT6UuZo__object_world_motion_direction__01", "video_id": "uvo__1PQcIT6UuZo", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (cyclist in black shirt and shorts) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "up", "down", "right", "forward"], "frame_paths": ["media/9d3349aa4920be212891/F0.jpg", "media/9d3349aa4920be212891/F1.jpg", "media/9d3349aa4920be212891/F2.jpg", "media/9d3349aa4920be212891/F3.jpg", "media/9d3349aa4920be212891/F4.jpg", "media/9d3349aa4920be212891/F5.jpg", "media/9d3349aa4920be212891/F6.jpg", "media/9d3349aa4920be212891/F7.jpg", "media/9d3349aa4920be212891/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (cyclist in black shirt and shorts) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. up\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8b98f395c3afd58c0a1e1b3049f9e0be8be2bbb9b3a99561ca08a76620cc5d2e"}
{"qa_id": "uvo__1PQcIT6UuZo__camera_object_distance_change__02", "video_id": "uvo__1PQcIT6UuZo", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did object A (cyclist in black shirt and shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9d3349aa4920be212891/F0.jpg", "media/9d3349aa4920be212891/F1.jpg", "media/9d3349aa4920be212891/F2.jpg", "media/9d3349aa4920be212891/F3.jpg", "media/9d3349aa4920be212891/F4.jpg", "media/9d3349aa4920be212891/F5.jpg", "media/9d3349aa4920be212891/F6.jpg", "media/9d3349aa4920be212891/F7.jpg", "media/9d3349aa4920be212891/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (cyclist in black shirt and shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40ec267a888a20dbcea3f46e3f50493473173f175f0f04c7d3f6aa506b53e4a1"}
{"qa_id": "uvo__1PXBrBUOmSI__object_world_motion_direction__00", "video_id": "uvo__1PXBrBUOmSI", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in dark clothing walking near the stream) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "down", "right", "forward", "backward"], "frame_paths": ["media/2b36c68dc4a0008066d7/F0.jpg", "media/2b36c68dc4a0008066d7/F1.jpg", "media/2b36c68dc4a0008066d7/F2.jpg", "media/2b36c68dc4a0008066d7/F3.jpg", "media/2b36c68dc4a0008066d7/F4.jpg", "media/2b36c68dc4a0008066d7/F5.jpg", "media/2b36c68dc4a0008066d7/F6.jpg", "media/2b36c68dc4a0008066d7/F7.jpg", "media/2b36c68dc4a0008066d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in dark clothing walking near the stream) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. down\nD. right\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1197c081e5481e8dc6e11b91c6b5e5465fb8881ba06192cd72fd5876ad07fe17"}
{"qa_id": "uvo__1RDhbqybg_I__camera_view_direction_change__00", "video_id": "uvo__1RDhbqybg_I", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/53dd871f5e463293ec71/F0.jpg", "media/53dd871f5e463293ec71/F1.jpg", "media/53dd871f5e463293ec71/F2.jpg", "media/53dd871f5e463293ec71/F3.jpg", "media/53dd871f5e463293ec71/F4.jpg", "media/53dd871f5e463293ec71/F5.jpg", "media/53dd871f5e463293ec71/F6.jpg", "media/53dd871f5e463293ec71/F7.jpg", "media/53dd871f5e463293ec71/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16a4ca6eeb0d84a0d96a1b095761436c4777b97628640bbab69e354a88eb5d55"}
{"qa_id": "uvo__1RWLlXFPWk4__camera_view_direction_change__00", "video_id": "uvo__1RWLlXFPWk4", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/3b96e7409fbe8346957f/F0.jpg", "media/3b96e7409fbe8346957f/F1.jpg", "media/3b96e7409fbe8346957f/F2.jpg", "media/3b96e7409fbe8346957f/F3.jpg", "media/3b96e7409fbe8346957f/F4.jpg", "media/3b96e7409fbe8346957f/F5.jpg", "media/3b96e7409fbe8346957f/F6.jpg", "media/3b96e7409fbe8346957f/F7.jpg", "media/3b96e7409fbe8346957f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "uvo__1SIQD7ocPM4__camera_view_direction_change__00", "video_id": "uvo__1SIQD7ocPM4", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/9fbd891d931e7d699222/F0.jpg", "media/9fbd891d931e7d699222/F1.jpg", "media/9fbd891d931e7d699222/F2.jpg", "media/9fbd891d931e7d699222/F3.jpg", "media/9fbd891d931e7d699222/F4.jpg", "media/9fbd891d931e7d699222/F5.jpg", "media/9fbd891d931e7d699222/F6.jpg", "media/9fbd891d931e7d699222/F7.jpg", "media/9fbd891d931e7d699222/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f9558cfcb39b4b8062b0685a5042b148145270b4c4b58790fc901bc14fc2ec8"}
{"qa_id": "uvo__1SIQD7ocPM4__camera_object_distance_change__01", "video_id": "uvo__1SIQD7ocPM4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (young boy with sunglasses and patterned swim trunks) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/660911e652abc7df0c70/F0.jpg", "media/660911e652abc7df0c70/F1.jpg", "media/660911e652abc7df0c70/F2.jpg", "media/660911e652abc7df0c70/F3.jpg", "media/660911e652abc7df0c70/F4.jpg", "media/660911e652abc7df0c70/F5.jpg", "media/660911e652abc7df0c70/F6.jpg", "media/660911e652abc7df0c70/F7.jpg", "media/660911e652abc7df0c70/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (young boy with sunglasses and patterned swim trunks) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fc8142929f77a97b590bfd0ea195b09d6b9b4c56e5637b4734d552afad18cad"}
{"qa_id": "uvo__1SscAwdPasM__object_world_motion_direction__00", "video_id": "uvo__1SscAwdPasM", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in red jacket and black pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "left", "right", "down", "up"], "frame_paths": ["media/9107ae7bfa48bb81ed3c/F0.jpg", "media/9107ae7bfa48bb81ed3c/F1.jpg", "media/9107ae7bfa48bb81ed3c/F2.jpg", "media/9107ae7bfa48bb81ed3c/F3.jpg", "media/9107ae7bfa48bb81ed3c/F4.jpg", "media/9107ae7bfa48bb81ed3c/F5.jpg", "media/9107ae7bfa48bb81ed3c/F6.jpg", "media/9107ae7bfa48bb81ed3c/F7.jpg", "media/9107ae7bfa48bb81ed3c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in red jacket and black pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. left\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "783f3e5083406909580e67eeb39cb3fef4642cb2af9d1d48e58e1341e0072f92"}
{"qa_id": "uvo__1SscAwdPasM__camera_object_distance_change__01", "video_id": "uvo__1SscAwdPasM", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F8, did object A (person in red jacket and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9107ae7bfa48bb81ed3c/F0.jpg", "media/9107ae7bfa48bb81ed3c/F1.jpg", "media/9107ae7bfa48bb81ed3c/F2.jpg", "media/9107ae7bfa48bb81ed3c/F3.jpg", "media/9107ae7bfa48bb81ed3c/F4.jpg", "media/9107ae7bfa48bb81ed3c/F5.jpg", "media/9107ae7bfa48bb81ed3c/F6.jpg", "media/9107ae7bfa48bb81ed3c/F7.jpg", "media/9107ae7bfa48bb81ed3c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (person in red jacket and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f4bc5a27200be4429717e06057ffeebeb95d9d65ea9e7b2b1fe87f52d413d029"}
{"qa_id": "uvo__1U3F4rtpMHs__camera_translation_direction__00", "video_id": "uvo__1U3F4rtpMHs", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "up", "backward", "forward"], "frame_paths": ["media/5e21c56355e3b3049b87/F0.jpg", "media/5e21c56355e3b3049b87/F1.jpg", "media/5e21c56355e3b3049b87/F2.jpg", "media/5e21c56355e3b3049b87/F3.jpg", "media/5e21c56355e3b3049b87/F4.jpg", "media/5e21c56355e3b3049b87/F5.jpg", "media/5e21c56355e3b3049b87/F6.jpg", "media/5e21c56355e3b3049b87/F7.jpg", "media/5e21c56355e3b3049b87/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. up\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "828b9b8c46ed6ffb5789b982695b64a9c58143e5dfa03bcb43cfc0ec67714d16"}
{"qa_id": "uvo__1U3F4rtpMHs__object_world_motion_direction__02", "video_id": "uvo__1U3F4rtpMHs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object B (brown horse with a rider wearing a light blue shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "up", "forward", "backward"], "frame_paths": ["media/673127b34874c13c9d16/F0.jpg", "media/673127b34874c13c9d16/F1.jpg", "media/673127b34874c13c9d16/F2.jpg", "media/673127b34874c13c9d16/F3.jpg", "media/673127b34874c13c9d16/F4.jpg", "media/673127b34874c13c9d16/F5.jpg", "media/673127b34874c13c9d16/F6.jpg", "media/673127b34874c13c9d16/F7.jpg", "media/673127b34874c13c9d16/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object B (brown horse with a rider wearing a light blue shirt) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5cef3ef0c7745c62d37b47eef306221cd0d92ff7b67aa08161bc1e8ab78dc7e"}
{"qa_id": "uvo__1U3F4rtpMHs__object_pair_distance_change__03", "video_id": "uvo__1U3F4rtpMHs", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, did the 3D distance between object A (horse with large ears seen from a first-person perspective) and object B (brown horse with a rider wearing a light blue shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/e47ba30218ffe4d483ef/F0.jpg", "media/e47ba30218ffe4d483ef/F1.jpg", "media/e47ba30218ffe4d483ef/F2.jpg", "media/e47ba30218ffe4d483ef/F3.jpg", "media/e47ba30218ffe4d483ef/F4.jpg", "media/e47ba30218ffe4d483ef/F5.jpg", "media/e47ba30218ffe4d483ef/F6.jpg", "media/e47ba30218ffe4d483ef/F7.jpg", "media/e47ba30218ffe4d483ef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (horse with large ears seen from a first-person perspective) and object B (brown horse with a rider wearing a light blue shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fcf2b2e45624c6b218352bbbfc78d0f4bcf1d4a148da3929842a486d49181884"}
{"qa_id": "uvo__1WMulo84kBY__camera_view_direction_change__00", "video_id": "uvo__1WMulo84kBY", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/85bd4f11fdc0eaa66849/F0.jpg", "media/85bd4f11fdc0eaa66849/F1.jpg", "media/85bd4f11fdc0eaa66849/F2.jpg", "media/85bd4f11fdc0eaa66849/F3.jpg", "media/85bd4f11fdc0eaa66849/F4.jpg", "media/85bd4f11fdc0eaa66849/F5.jpg", "media/85bd4f11fdc0eaa66849/F6.jpg", "media/85bd4f11fdc0eaa66849/F7.jpg", "media/85bd4f11fdc0eaa66849/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654016ae105affb53bc2face594c67e96f587f4cdc806a6f418076b30032d255"}
{"qa_id": "uvo__1WOies49wDc__object_world_motion_direction__00", "video_id": "uvo__1WOies49wDc", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object A (green and yellow striped parasail canopy) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "up", "forward"], "frame_paths": ["media/d3558f9510b5927f2087/F0.jpg", "media/d3558f9510b5927f2087/F1.jpg", "media/d3558f9510b5927f2087/F2.jpg", "media/d3558f9510b5927f2087/F3.jpg", "media/d3558f9510b5927f2087/F4.jpg", "media/d3558f9510b5927f2087/F5.jpg", "media/d3558f9510b5927f2087/F6.jpg", "media/d3558f9510b5927f2087/F7.jpg", "media/d3558f9510b5927f2087/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object A (green and yellow striped parasail canopy) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fc0e2ca03a2bfc6481b280b1e0b7b4e800bcda54f4be313995e7cb8579c0c8f3"}
{"qa_id": "uvo__1WOies49wDc__camera_object_distance_change__01", "video_id": "uvo__1WOies49wDc", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object A (green and yellow striped parasail canopy) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d3558f9510b5927f2087/F0.jpg", "media/d3558f9510b5927f2087/F1.jpg", "media/d3558f9510b5927f2087/F2.jpg", "media/d3558f9510b5927f2087/F3.jpg", "media/d3558f9510b5927f2087/F4.jpg", "media/d3558f9510b5927f2087/F5.jpg", "media/d3558f9510b5927f2087/F6.jpg", "media/d3558f9510b5927f2087/F7.jpg", "media/d3558f9510b5927f2087/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (green and yellow striped parasail canopy) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44e4888e61df5f07b32498bb7ad3271f8cc1d0bbfe4aca4fe29703bdcda2d86a"}
{"qa_id": "uvo__1_97vo5Mo0Q__camera_view_direction_change__00", "video_id": "uvo__1_97vo5Mo0Q", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/ef4e77c53d46a1f75b27/F0.jpg", "media/ef4e77c53d46a1f75b27/F1.jpg", "media/ef4e77c53d46a1f75b27/F2.jpg", "media/ef4e77c53d46a1f75b27/F3.jpg", "media/ef4e77c53d46a1f75b27/F4.jpg", "media/ef4e77c53d46a1f75b27/F5.jpg", "media/ef4e77c53d46a1f75b27/F6.jpg", "media/ef4e77c53d46a1f75b27/F7.jpg", "media/ef4e77c53d46a1f75b27/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "uvo__1_mioTBdjgM__camera_view_direction_change__00", "video_id": "uvo__1_mioTBdjgM", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/65a686c6ba05156e251b/F0.jpg", "media/65a686c6ba05156e251b/F1.jpg", "media/65a686c6ba05156e251b/F2.jpg", "media/65a686c6ba05156e251b/F3.jpg", "media/65a686c6ba05156e251b/F4.jpg", "media/65a686c6ba05156e251b/F5.jpg", "media/65a686c6ba05156e251b/F6.jpg", "media/65a686c6ba05156e251b/F7.jpg", "media/65a686c6ba05156e251b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1d4c0549ccb65dde198f94b7b700a39260affcc8a68eef0813868681f0dd4466"}
{"qa_id": "uvo__1aA7lPPfCrw__camera_object_distance_change__00", "video_id": "uvo__1aA7lPPfCrw", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did object A (person in white shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/6eacf8516a13fe347d51/F0.jpg", "media/6eacf8516a13fe347d51/F1.jpg", "media/6eacf8516a13fe347d51/F2.jpg", "media/6eacf8516a13fe347d51/F3.jpg", "media/6eacf8516a13fe347d51/F4.jpg", "media/6eacf8516a13fe347d51/F5.jpg", "media/6eacf8516a13fe347d51/F6.jpg", "media/6eacf8516a13fe347d51/F7.jpg", "media/6eacf8516a13fe347d51/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (person in white shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a1efd85be247ff1b5c8d00ad58f4564235592c852755c7df71ce734407f5ad00"}
{"qa_id": "uvo__1cLg-P8GLOc__camera_view_direction_change__00", "video_id": "uvo__1cLg-P8GLOc", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/9a40b4edad216d2263e2/F0.jpg", "media/9a40b4edad216d2263e2/F1.jpg", "media/9a40b4edad216d2263e2/F2.jpg", "media/9a40b4edad216d2263e2/F3.jpg", "media/9a40b4edad216d2263e2/F4.jpg", "media/9a40b4edad216d2263e2/F5.jpg", "media/9a40b4edad216d2263e2/F6.jpg", "media/9a40b4edad216d2263e2/F7.jpg", "media/9a40b4edad216d2263e2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "uvo__1dRL83xXz4M__camera_translation_direction__00", "video_id": "uvo__1dRL83xXz4M", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "backward", "forward", "left", "right"], "frame_paths": ["media/fb91feb865ce674ef070/F0.jpg", "media/fb91feb865ce674ef070/F1.jpg", "media/fb91feb865ce674ef070/F2.jpg", "media/fb91feb865ce674ef070/F3.jpg", "media/fb91feb865ce674ef070/F4.jpg", "media/fb91feb865ce674ef070/F5.jpg", "media/fb91feb865ce674ef070/F6.jpg", "media/fb91feb865ce674ef070/F7.jpg", "media/fb91feb865ce674ef070/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. backward\nD. forward\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "91e6aed7d804ddb4046708c706b86fe20c43c4ac24299eb729987a15fe92d66b"}
{"qa_id": "uvo__1dRL83xXz4M__camera_view_direction_change__01", "video_id": "uvo__1dRL83xXz4M", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/fb91feb865ce674ef070/F0.jpg", "media/fb91feb865ce674ef070/F1.jpg", "media/fb91feb865ce674ef070/F2.jpg", "media/fb91feb865ce674ef070/F3.jpg", "media/fb91feb865ce674ef070/F4.jpg", "media/fb91feb865ce674ef070/F5.jpg", "media/fb91feb865ce674ef070/F6.jpg", "media/fb91feb865ce674ef070/F7.jpg", "media/fb91feb865ce674ef070/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6eae66ee8c9fd961541d1dbc1fbfab09c9704c7eed5843b6a680b6012297addc"}
{"qa_id": "uvo__1damFpFAznY__object_world_motion_direction__00", "video_id": "uvo__1damFpFAznY", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (man in blue and white shirt holding a long pole) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "up", "forward", "right"], "frame_paths": ["media/c850e83803b578965300/F0.jpg", "media/c850e83803b578965300/F1.jpg", "media/c850e83803b578965300/F2.jpg", "media/c850e83803b578965300/F3.jpg", "media/c850e83803b578965300/F4.jpg", "media/c850e83803b578965300/F5.jpg", "media/c850e83803b578965300/F6.jpg", "media/c850e83803b578965300/F7.jpg", "media/c850e83803b578965300/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (man in blue and white shirt holding a long pole) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1dbd366c6545aba6771a04229940a18ae753c4605cccc1b09d872f463ee268d2"}
{"qa_id": "uvo__1damFpFAznY__camera_object_distance_change__01", "video_id": "uvo__1damFpFAznY", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (man in blue and white shirt holding a long pole) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/c850e83803b578965300/F0.jpg", "media/c850e83803b578965300/F1.jpg", "media/c850e83803b578965300/F2.jpg", "media/c850e83803b578965300/F3.jpg", "media/c850e83803b578965300/F4.jpg", "media/c850e83803b578965300/F5.jpg", "media/c850e83803b578965300/F6.jpg", "media/c850e83803b578965300/F7.jpg", "media/c850e83803b578965300/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (man in blue and white shirt holding a long pole) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b60c5499b9770f22f155b510e7c83318f7c34ef2ca28d46483f6171cdcec4d6"}
{"qa_id": "uvo__1damFpFAznY__object_world_motion_direction__02", "video_id": "uvo__1damFpFAznY", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (man in white shirt and dark shorts walking on the track) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "right", "left", "down", "backward", "forward"], "frame_paths": ["media/b7139a8f68debe990e30/F0.jpg", "media/b7139a8f68debe990e30/F1.jpg", "media/b7139a8f68debe990e30/F2.jpg", "media/b7139a8f68debe990e30/F3.jpg", "media/b7139a8f68debe990e30/F4.jpg", "media/b7139a8f68debe990e30/F5.jpg", "media/b7139a8f68debe990e30/F6.jpg", "media/b7139a8f68debe990e30/F7.jpg", "media/b7139a8f68debe990e30/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (man in white shirt and dark shorts walking on the track) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. left\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "db1a8601fc8630f9e49e6a681d09a5aee72f8267b3bc648f1c13a3e3691f33b7"}
{"qa_id": "uvo__1damFpFAznY__object_pair_distance_change__03", "video_id": "uvo__1damFpFAznY", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did the 3D distance between object A (man in blue and white shirt holding a long pole) and object B (man in white shirt and dark shorts walking on the track) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/f8b50c5afca2086e768d/F0.jpg", "media/f8b50c5afca2086e768d/F1.jpg", "media/f8b50c5afca2086e768d/F2.jpg", "media/f8b50c5afca2086e768d/F3.jpg", "media/f8b50c5afca2086e768d/F4.jpg", "media/f8b50c5afca2086e768d/F5.jpg", "media/f8b50c5afca2086e768d/F6.jpg", "media/f8b50c5afca2086e768d/F7.jpg", "media/f8b50c5afca2086e768d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (man in blue and white shirt holding a long pole) and object B (man in white shirt and dark shorts walking on the track) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "37175dc63c3c37a512a0e97fff3ddb1fd7150eb990b60a558f4f9b233b6e738e"}
{"qa_id": "uvo__1dmwnxQIC0I__camera_view_direction_change__00", "video_id": "uvo__1dmwnxQIC0I", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/e54407da814220c03142/F0.jpg", "media/e54407da814220c03142/F1.jpg", "media/e54407da814220c03142/F2.jpg", "media/e54407da814220c03142/F3.jpg", "media/e54407da814220c03142/F4.jpg", "media/e54407da814220c03142/F5.jpg", "media/e54407da814220c03142/F6.jpg", "media/e54407da814220c03142/F7.jpg", "media/e54407da814220c03142/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7fd29d446bf5398df4a2b54fd51c532ab21432d6e315f4f25ed8f1cfa69f9b6"}
{"qa_id": "uvo__1dvRF5N2BIs__object_world_motion_direction__00", "video_id": "uvo__1dvRF5N2BIs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (skier in dark clothing) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "forward", "left", "up", "right"], "frame_paths": ["media/96591cdccb0c92249e8f/F0.jpg", "media/96591cdccb0c92249e8f/F1.jpg", "media/96591cdccb0c92249e8f/F2.jpg", "media/96591cdccb0c92249e8f/F3.jpg", "media/96591cdccb0c92249e8f/F4.jpg", "media/96591cdccb0c92249e8f/F5.jpg", "media/96591cdccb0c92249e8f/F6.jpg", "media/96591cdccb0c92249e8f/F7.jpg", "media/96591cdccb0c92249e8f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (skier in dark clothing) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. forward\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1df0da794070c822eabce2373fae4c4553c78f5c3e9fa3cec74f1c9efaf53932"}
{"qa_id": "uvo__1dvRF5N2BIs__camera_object_distance_change__01", "video_id": "uvo__1dvRF5N2BIs", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (skier in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/96591cdccb0c92249e8f/F0.jpg", "media/96591cdccb0c92249e8f/F1.jpg", "media/96591cdccb0c92249e8f/F2.jpg", "media/96591cdccb0c92249e8f/F3.jpg", "media/96591cdccb0c92249e8f/F4.jpg", "media/96591cdccb0c92249e8f/F5.jpg", "media/96591cdccb0c92249e8f/F6.jpg", "media/96591cdccb0c92249e8f/F7.jpg", "media/96591cdccb0c92249e8f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (skier in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "606d3e01d281a6245e69a0f6ba3b4dc1422f06cc1d3ba9e01d22ebc1f76a9e1c"}
{"qa_id": "uvo__1gRJ-gl2_2A__camera_view_direction_change__00", "video_id": "uvo__1gRJ-gl2_2A", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/2d3cc7110e54fd19e563/F0.jpg", "media/2d3cc7110e54fd19e563/F1.jpg", "media/2d3cc7110e54fd19e563/F2.jpg", "media/2d3cc7110e54fd19e563/F3.jpg", "media/2d3cc7110e54fd19e563/F4.jpg", "media/2d3cc7110e54fd19e563/F5.jpg", "media/2d3cc7110e54fd19e563/F6.jpg", "media/2d3cc7110e54fd19e563/F7.jpg", "media/2d3cc7110e54fd19e563/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "93fd8df73d31bc8e34ac6bdb52cfdf36ae7d1d71294b0a1e86c1abe14ef61687"}
{"qa_id": "uvo__1gUai0f0Y18__camera_view_direction_change__00", "video_id": "uvo__1gUai0f0Y18", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/ba4730cb27a18a798498/F0.jpg", "media/ba4730cb27a18a798498/F1.jpg", "media/ba4730cb27a18a798498/F2.jpg", "media/ba4730cb27a18a798498/F3.jpg", "media/ba4730cb27a18a798498/F4.jpg", "media/ba4730cb27a18a798498/F5.jpg", "media/ba4730cb27a18a798498/F6.jpg", "media/ba4730cb27a18a798498/F7.jpg", "media/ba4730cb27a18a798498/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "139d78c609346a36b31a51baa6a9cfbf5b99d316434cbd3d4bd9ef34f83af385"}
{"qa_id": "uvo__1gUai0f0Y18__camera_object_distance_change__02", "video_id": "uvo__1gUai0f0Y18", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did object A (person in black shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/239f8bc2e6693445c61b/F0.jpg", "media/239f8bc2e6693445c61b/F1.jpg", "media/239f8bc2e6693445c61b/F2.jpg", "media/239f8bc2e6693445c61b/F3.jpg", "media/239f8bc2e6693445c61b/F4.jpg", "media/239f8bc2e6693445c61b/F5.jpg", "media/239f8bc2e6693445c61b/F6.jpg", "media/239f8bc2e6693445c61b/F7.jpg", "media/239f8bc2e6693445c61b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (person in black shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5e77ac5e3a7b8ae608671a130a053d4fa87b14a6288fe00abacc18d8220febde"}
{"qa_id": "uvo__1gUai0f0Y18__object_pair_distance_change__03", "video_id": "uvo__1gUai0f0Y18", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did the 3D distance between object A (person in black shirt) and object B (person in grey hoodie) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2f988e2a57ecaa4ec640/F0.jpg", "media/2f988e2a57ecaa4ec640/F1.jpg", "media/2f988e2a57ecaa4ec640/F2.jpg", "media/2f988e2a57ecaa4ec640/F3.jpg", "media/2f988e2a57ecaa4ec640/F4.jpg", "media/2f988e2a57ecaa4ec640/F5.jpg", "media/2f988e2a57ecaa4ec640/F6.jpg", "media/2f988e2a57ecaa4ec640/F7.jpg", "media/2f988e2a57ecaa4ec640/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (person in black shirt) and object B (person in grey hoodie) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9ee09fcf86a6d79ce615604852af3c313e42dc59ce7641a66b3f2ac58619ae0c"}
{"qa_id": "uvo__1gUai0f0Y18__object_travel_distance_comparison__04", "video_id": "uvo__1gUai0f0Y18", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black shirt) or B (person in grey hoodie)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2f988e2a57ecaa4ec640/F0.jpg", "media/2f988e2a57ecaa4ec640/F1.jpg", "media/2f988e2a57ecaa4ec640/F2.jpg", "media/2f988e2a57ecaa4ec640/F3.jpg", "media/2f988e2a57ecaa4ec640/F4.jpg", "media/2f988e2a57ecaa4ec640/F5.jpg", "media/2f988e2a57ecaa4ec640/F6.jpg", "media/2f988e2a57ecaa4ec640/F7.jpg", "media/2f988e2a57ecaa4ec640/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in black shirt) or B (person in grey hoodie)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7b76d7a6aca5134a684e4ee6f71314c0a6c8a990819d83e0257f6b0d9a425d70"}
{"qa_id": "uvo__1k_T1EBuCSs__camera_translation_direction__00", "video_id": "uvo__1k_T1EBuCSs", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "backward", "up", "forward"], "frame_paths": ["media/346d18cc5b1e81861518/F0.jpg", "media/346d18cc5b1e81861518/F1.jpg", "media/346d18cc5b1e81861518/F2.jpg", "media/346d18cc5b1e81861518/F3.jpg", "media/346d18cc5b1e81861518/F4.jpg", "media/346d18cc5b1e81861518/F5.jpg", "media/346d18cc5b1e81861518/F6.jpg", "media/346d18cc5b1e81861518/F7.jpg", "media/346d18cc5b1e81861518/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e9c2858cae3bdd983a9f9fef7637d685c70325a5179d614733bf4d8faf73f4b"}
{"qa_id": "uvo__1l84CF_6o5w__camera_translation_direction__00", "video_id": "uvo__1l84CF_6o5w", "task": "camera_translation_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "left", "backward", "down", "up"], "frame_paths": ["media/f1d39121537c046edd7a/F0.jpg", "media/f1d39121537c046edd7a/F1.jpg", "media/f1d39121537c046edd7a/F2.jpg", "media/f1d39121537c046edd7a/F3.jpg", "media/f1d39121537c046edd7a/F4.jpg", "media/f1d39121537c046edd7a/F5.jpg", "media/f1d39121537c046edd7a/F6.jpg", "media/f1d39121537c046edd7a/F7.jpg", "media/f1d39121537c046edd7a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. left\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5e6829967cc08a74dcda93d75202287effeef930d209ecdb0e6f7cca58c7e2b"}
{"qa_id": "uvo__1l84CF_6o5w__camera_view_direction_change__01", "video_id": "uvo__1l84CF_6o5w", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/f1d39121537c046edd7a/F0.jpg", "media/f1d39121537c046edd7a/F1.jpg", "media/f1d39121537c046edd7a/F2.jpg", "media/f1d39121537c046edd7a/F3.jpg", "media/f1d39121537c046edd7a/F4.jpg", "media/f1d39121537c046edd7a/F5.jpg", "media/f1d39121537c046edd7a/F6.jpg", "media/f1d39121537c046edd7a/F7.jpg", "media/f1d39121537c046edd7a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "603de446c5814db82051c70f8311b3a130eb38a2dbf02023686fab05c41c4c76"}
{"qa_id": "uvo__1l84CF_6o5w__object_world_motion_direction__02", "video_id": "uvo__1l84CF_6o5w", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in dark jacket and orange pants riding a snowmobile) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "backward", "right"], "frame_paths": ["media/63d37db882b048250d32/F0.jpg", "media/63d37db882b048250d32/F1.jpg", "media/63d37db882b048250d32/F2.jpg", "media/63d37db882b048250d32/F3.jpg", "media/63d37db882b048250d32/F4.jpg", "media/63d37db882b048250d32/F5.jpg", "media/63d37db882b048250d32/F6.jpg", "media/63d37db882b048250d32/F7.jpg", "media/63d37db882b048250d32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in dark jacket and orange pants riding a snowmobile) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "535753c27eef5a314ec4f02be3d425aad297d1f92358bfc49645d9aa5bb03682"}
{"qa_id": "uvo__1l84CF_6o5w__camera_object_distance_change__03", "video_id": "uvo__1l84CF_6o5w", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object A (person in dark jacket and orange pants riding a snowmobile) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/63d37db882b048250d32/F0.jpg", "media/63d37db882b048250d32/F1.jpg", "media/63d37db882b048250d32/F2.jpg", "media/63d37db882b048250d32/F3.jpg", "media/63d37db882b048250d32/F4.jpg", "media/63d37db882b048250d32/F5.jpg", "media/63d37db882b048250d32/F6.jpg", "media/63d37db882b048250d32/F7.jpg", "media/63d37db882b048250d32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (person in dark jacket and orange pants riding a snowmobile) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6e894790638d7bb2dc51c99e734523338acdc29a4898a92baa989b20c77a4e80"}
{"qa_id": "uvo__1l84CF_6o5w__object_world_motion_direction__04", "video_id": "uvo__1l84CF_6o5w", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object B (black and yellow snowmobile) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "down", "backward", "right", "up"], "frame_paths": ["media/731f7a6c079d2566bb7c/F0.jpg", "media/731f7a6c079d2566bb7c/F1.jpg", "media/731f7a6c079d2566bb7c/F2.jpg", "media/731f7a6c079d2566bb7c/F3.jpg", "media/731f7a6c079d2566bb7c/F4.jpg", "media/731f7a6c079d2566bb7c/F5.jpg", "media/731f7a6c079d2566bb7c/F6.jpg", "media/731f7a6c079d2566bb7c/F7.jpg", "media/731f7a6c079d2566bb7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object B (black and yellow snowmobile) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. down\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd2e4f14b561305f09937062f8040d3beb07bafef281bf54b2b7cef028d96922"}
{"qa_id": "uvo__1mBPyRkZefY__camera_view_direction_change__00", "video_id": "uvo__1mBPyRkZefY", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/da902cc6201cfa798d05/F0.jpg", "media/da902cc6201cfa798d05/F1.jpg", "media/da902cc6201cfa798d05/F2.jpg", "media/da902cc6201cfa798d05/F3.jpg", "media/da902cc6201cfa798d05/F4.jpg", "media/da902cc6201cfa798d05/F5.jpg", "media/da902cc6201cfa798d05/F6.jpg", "media/da902cc6201cfa798d05/F7.jpg", "media/da902cc6201cfa798d05/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "uvo__1nVmIs3U_l8__camera_view_direction_change__00", "video_id": "uvo__1nVmIs3U_l8", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/db21693bf33e33564f3c/F0.jpg", "media/db21693bf33e33564f3c/F1.jpg", "media/db21693bf33e33564f3c/F2.jpg", "media/db21693bf33e33564f3c/F3.jpg", "media/db21693bf33e33564f3c/F4.jpg", "media/db21693bf33e33564f3c/F5.jpg", "media/db21693bf33e33564f3c/F6.jpg", "media/db21693bf33e33564f3c/F7.jpg", "media/db21693bf33e33564f3c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4d092697e7716cd1ba88194bcaa5c2ee56b76809ce4814605762fa830a476c49"}
{"qa_id": "uvo__1rRZZKnWCo8__camera_view_direction_change__00", "video_id": "uvo__1rRZZKnWCo8", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/f2b5538c1b35b195f86b/F0.jpg", "media/f2b5538c1b35b195f86b/F1.jpg", "media/f2b5538c1b35b195f86b/F2.jpg", "media/f2b5538c1b35b195f86b/F3.jpg", "media/f2b5538c1b35b195f86b/F4.jpg", "media/f2b5538c1b35b195f86b/F5.jpg", "media/f2b5538c1b35b195f86b/F6.jpg", "media/f2b5538c1b35b195f86b/F7.jpg", "media/f2b5538c1b35b195f86b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9557c0796e28ae2a4f32990ec2b747f1a5933b7979d70ae5cf47b924054e1483"}
{"qa_id": "uvo__1t3OlVUHTtY__object_world_motion_direction__00", "video_id": "uvo__1t3OlVUHTtY", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object B (person in the water) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "left", "right", "down"], "frame_paths": ["media/34d5d6ca7f2dca26c8a6/F0.jpg", "media/34d5d6ca7f2dca26c8a6/F1.jpg", "media/34d5d6ca7f2dca26c8a6/F2.jpg", "media/34d5d6ca7f2dca26c8a6/F3.jpg", "media/34d5d6ca7f2dca26c8a6/F4.jpg", "media/34d5d6ca7f2dca26c8a6/F5.jpg", "media/34d5d6ca7f2dca26c8a6/F6.jpg", "media/34d5d6ca7f2dca26c8a6/F7.jpg", "media/34d5d6ca7f2dca26c8a6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object B (person in the water) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5dfd20c5daac932435996b33fd0ae24484996d8f74cad914f73e25def0fa5c90"}
{"qa_id": "uvo__1zBfMrwszgc__object_world_motion_direction__00", "video_id": "uvo__1zBfMrwszgc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in red shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "down", "backward", "up", "left", "forward"], "frame_paths": ["media/e573e41b528bcf85c07d/F0.jpg", "media/e573e41b528bcf85c07d/F1.jpg", "media/e573e41b528bcf85c07d/F2.jpg", "media/e573e41b528bcf85c07d/F3.jpg", "media/e573e41b528bcf85c07d/F4.jpg", "media/e573e41b528bcf85c07d/F5.jpg", "media/e573e41b528bcf85c07d/F6.jpg", "media/e573e41b528bcf85c07d/F7.jpg", "media/e573e41b528bcf85c07d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person in red shirt) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. backward\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "805d94a8477b5442d0333b2bed09b3cf92bca73e2bc8e2b53899c49e29bea742"}
{"qa_id": "uvo__1zBfMrwszgc__object_world_motion_direction__01", "video_id": "uvo__1zBfMrwszgc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (person in black shirt with white text) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "left", "up", "backward", "down"], "frame_paths": ["media/c79cef68ce39c4a80d44/F0.jpg", "media/c79cef68ce39c4a80d44/F1.jpg", "media/c79cef68ce39c4a80d44/F2.jpg", "media/c79cef68ce39c4a80d44/F3.jpg", "media/c79cef68ce39c4a80d44/F4.jpg", "media/c79cef68ce39c4a80d44/F5.jpg", "media/c79cef68ce39c4a80d44/F6.jpg", "media/c79cef68ce39c4a80d44/F7.jpg", "media/c79cef68ce39c4a80d44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (person in black shirt with white text) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. left\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "266bdcb0942e9baede400074ceb160d2bff747483386a5391ac09a2a07a503e6"}
{"qa_id": "uvo__1zBfMrwszgc__camera_object_distance_change__02", "video_id": "uvo__1zBfMrwszgc", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object B (person in black shirt with white text) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c79cef68ce39c4a80d44/F0.jpg", "media/c79cef68ce39c4a80d44/F1.jpg", "media/c79cef68ce39c4a80d44/F2.jpg", "media/c79cef68ce39c4a80d44/F3.jpg", "media/c79cef68ce39c4a80d44/F4.jpg", "media/c79cef68ce39c4a80d44/F5.jpg", "media/c79cef68ce39c4a80d44/F6.jpg", "media/c79cef68ce39c4a80d44/F7.jpg", "media/c79cef68ce39c4a80d44/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object B (person in black shirt with white text) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3505628c759438cd32bd0f9c8d7ebd5e5356abcdafd47ad916481b91dc13a6f1"}
{"qa_id": "uvo__1zBfMrwszgc__object_travel_distance_comparison__03", "video_id": "uvo__1zBfMrwszgc", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt) or B (person in black shirt with white text)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/86374420ad701c5c96a1/F0.jpg", "media/86374420ad701c5c96a1/F1.jpg", "media/86374420ad701c5c96a1/F2.jpg", "media/86374420ad701c5c96a1/F3.jpg", "media/86374420ad701c5c96a1/F4.jpg", "media/86374420ad701c5c96a1/F5.jpg", "media/86374420ad701c5c96a1/F6.jpg", "media/86374420ad701c5c96a1/F7.jpg", "media/86374420ad701c5c96a1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt) or B (person in black shirt with white text)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "89bd409c88cbf5c2e76c18672229726a51db74e154641d581ceb33b8067a8928"}
{"qa_id": "uvo__2-KpBW7QNIs__object_pair_distance_change__00", "video_id": "uvo__2-KpBW7QNIs", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F8, did the 3D distance between object A (person in red shirt and orange head covering) and object B (person in red shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/2f10885643eec2fe2918/F0.jpg", "media/2f10885643eec2fe2918/F1.jpg", "media/2f10885643eec2fe2918/F2.jpg", "media/2f10885643eec2fe2918/F3.jpg", "media/2f10885643eec2fe2918/F4.jpg", "media/2f10885643eec2fe2918/F5.jpg", "media/2f10885643eec2fe2918/F6.jpg", "media/2f10885643eec2fe2918/F7.jpg", "media/2f10885643eec2fe2918/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g6"}, "requested_object_ids": ["g1", "g6"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did the 3D distance between object A (person in red shirt and orange head covering) and object B (person in red shirt and black shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4927bd65fdd6cde2951bd005129b73252ddb564479af5e72fee2183ba6adac29"}
{"qa_id": "uvo__21m7Y8hto2o__camera_translation_direction__00", "video_id": "uvo__21m7Y8hto2o", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "backward", "forward", "right"], "frame_paths": ["media/8139d332035967bed0eb/F0.jpg", "media/8139d332035967bed0eb/F1.jpg", "media/8139d332035967bed0eb/F2.jpg", "media/8139d332035967bed0eb/F3.jpg", "media/8139d332035967bed0eb/F4.jpg", "media/8139d332035967bed0eb/F5.jpg", "media/8139d332035967bed0eb/F6.jpg", "media/8139d332035967bed0eb/F7.jpg", "media/8139d332035967bed0eb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. backward\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0fc6b35913eba82ac78b47944fcb9732ae7814043dc71849b3624c787a633e10"}
{"qa_id": "uvo__21m7Y8hto2o__object_world_motion_direction__01", "video_id": "uvo__21m7Y8hto2o", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in black jacket and orange pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "backward", "up", "forward"], "frame_paths": ["media/52a8af248e53cfb7d640/F0.jpg", "media/52a8af248e53cfb7d640/F1.jpg", "media/52a8af248e53cfb7d640/F2.jpg", "media/52a8af248e53cfb7d640/F3.jpg", "media/52a8af248e53cfb7d640/F4.jpg", "media/52a8af248e53cfb7d640/F5.jpg", "media/52a8af248e53cfb7d640/F6.jpg", "media/52a8af248e53cfb7d640/F7.jpg", "media/52a8af248e53cfb7d640/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in black jacket and orange pants) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65206b90948859529c5e5ae07dd12d5fed5222d0eeb57c5c4a1433d8247a6562"}
{"qa_id": "uvo__276U2IfXU6s__camera_view_direction_change__00", "video_id": "uvo__276U2IfXU6s", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/bf55f0b10a4161bafa1e/F0.jpg", "media/bf55f0b10a4161bafa1e/F1.jpg", "media/bf55f0b10a4161bafa1e/F2.jpg", "media/bf55f0b10a4161bafa1e/F3.jpg", "media/bf55f0b10a4161bafa1e/F4.jpg", "media/bf55f0b10a4161bafa1e/F5.jpg", "media/bf55f0b10a4161bafa1e/F6.jpg", "media/bf55f0b10a4161bafa1e/F7.jpg", "media/bf55f0b10a4161bafa1e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9ba2cc51b29405424f7d48f9d823ee5cd59f07f17efd65cd4610ac1ec100a34d"}
{"qa_id": "uvo__2B8s6prCDIg__object_travel_distance_comparison__00", "video_id": "uvo__2B8s6prCDIg", "task": "object_travel_distance_comparison", "group": "LLL", "group4": "LLL", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a red cap and black harness vest) or B (green and red paraglider canopy in the sky)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/6b1c6f0b33af8873dafa/F0.jpg", "media/6b1c6f0b33af8873dafa/F1.jpg", "media/6b1c6f0b33af8873dafa/F2.jpg", "media/6b1c6f0b33af8873dafa/F3.jpg", "media/6b1c6f0b33af8873dafa/F4.jpg", "media/6b1c6f0b33af8873dafa/F5.jpg", "media/6b1c6f0b33af8873dafa/F6.jpg", "media/6b1c6f0b33af8873dafa/F7.jpg", "media/6b1c6f0b33af8873dafa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person wearing a red cap and black harness vest) or B (green and red paraglider canopy in the sky)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a69c1ec01a6d564fec181190d9c747134bb64bb143617ba982219133d39af6a"}
{"qa_id": "uvo__2Bmjw4OGE2U__camera_view_direction_change__00", "video_id": "uvo__2Bmjw4OGE2U", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/0b550519246e7535f96c/F0.jpg", "media/0b550519246e7535f96c/F1.jpg", "media/0b550519246e7535f96c/F2.jpg", "media/0b550519246e7535f96c/F3.jpg", "media/0b550519246e7535f96c/F4.jpg", "media/0b550519246e7535f96c/F5.jpg", "media/0b550519246e7535f96c/F6.jpg", "media/0b550519246e7535f96c/F7.jpg", "media/0b550519246e7535f96c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "603de446c5814db82051c70f8311b3a130eb38a2dbf02023686fab05c41c4c76"}
{"qa_id": "uvo__2CG7MhTIdNc__object_world_motion_direction__00", "video_id": "uvo__2CG7MhTIdNc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (small child wearing a green t-shirt and dark shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "up", "right", "left"], "frame_paths": ["media/5f843caba335ed7ab35d/F0.jpg", "media/5f843caba335ed7ab35d/F1.jpg", "media/5f843caba335ed7ab35d/F2.jpg", "media/5f843caba335ed7ab35d/F3.jpg", "media/5f843caba335ed7ab35d/F4.jpg", "media/5f843caba335ed7ab35d/F5.jpg", "media/5f843caba335ed7ab35d/F6.jpg", "media/5f843caba335ed7ab35d/F7.jpg", "media/5f843caba335ed7ab35d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (small child wearing a green t-shirt and dark shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. up\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "12fe1f44ee51206055eba0000ba595d170dacd0ae2265c0de14478953b1442ef"}
{"qa_id": "uvo__2CG7MhTIdNc__object_pair_distance_change__01", "video_id": "uvo__2CG7MhTIdNc", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did the 3D distance between object A (tall person wearing a light blue t-shirt and dark shorts) and object B (small child wearing a green t-shirt and dark shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/a1c1aba51bbddb7834dd/F0.jpg", "media/a1c1aba51bbddb7834dd/F1.jpg", "media/a1c1aba51bbddb7834dd/F2.jpg", "media/a1c1aba51bbddb7834dd/F3.jpg", "media/a1c1aba51bbddb7834dd/F4.jpg", "media/a1c1aba51bbddb7834dd/F5.jpg", "media/a1c1aba51bbddb7834dd/F6.jpg", "media/a1c1aba51bbddb7834dd/F7.jpg", "media/a1c1aba51bbddb7834dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (tall person wearing a light blue t-shirt and dark shorts) and object B (small child wearing a green t-shirt and dark shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce91ac76406e77298f6104c83f37e2e154233ca8d67afde3c919f43b4faf4b94"}
{"qa_id": "uvo__2CG7MhTIdNc__object_travel_distance_comparison__02", "video_id": "uvo__2CG7MhTIdNc", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tall person wearing a light blue t-shirt and dark shorts) or B (small child wearing a green t-shirt and dark shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/a1c1aba51bbddb7834dd/F0.jpg", "media/a1c1aba51bbddb7834dd/F1.jpg", "media/a1c1aba51bbddb7834dd/F2.jpg", "media/a1c1aba51bbddb7834dd/F3.jpg", "media/a1c1aba51bbddb7834dd/F4.jpg", "media/a1c1aba51bbddb7834dd/F5.jpg", "media/a1c1aba51bbddb7834dd/F6.jpg", "media/a1c1aba51bbddb7834dd/F7.jpg", "media/a1c1aba51bbddb7834dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": ["g4", "g5"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F2 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (tall person wearing a light blue t-shirt and dark shorts) or B (small child wearing a green t-shirt and dark shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7ac864776fd48ad4e4642dd7c55d17202d6fa5242f568e71a0adb500c63f3887"}
{"qa_id": "uvo__2E0KINZFSHI__camera_view_direction_change__00", "video_id": "uvo__2E0KINZFSHI", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/ed2ea720b6101f47e253/F0.jpg", "media/ed2ea720b6101f47e253/F1.jpg", "media/ed2ea720b6101f47e253/F2.jpg", "media/ed2ea720b6101f47e253/F3.jpg", "media/ed2ea720b6101f47e253/F4.jpg", "media/ed2ea720b6101f47e253/F5.jpg", "media/ed2ea720b6101f47e253/F6.jpg", "media/ed2ea720b6101f47e253/F7.jpg", "media/ed2ea720b6101f47e253/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "116f284bbcff9f3ff536439744d342e78b21fb4c348e069dc494a10fb955ecc6"}
{"qa_id": "uvo__2ElBbDU9NpE__camera_view_direction_change__00", "video_id": "uvo__2ElBbDU9NpE", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/636edaf59e7a2c8e6add/F0.jpg", "media/636edaf59e7a2c8e6add/F1.jpg", "media/636edaf59e7a2c8e6add/F2.jpg", "media/636edaf59e7a2c8e6add/F3.jpg", "media/636edaf59e7a2c8e6add/F4.jpg", "media/636edaf59e7a2c8e6add/F5.jpg", "media/636edaf59e7a2c8e6add/F6.jpg", "media/636edaf59e7a2c8e6add/F7.jpg", "media/636edaf59e7a2c8e6add/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "12c5cad0216df609a82185774c2897728c0daac6ce8a8d410c8fff051efe06cc"}
{"qa_id": "uvo__2FGgpJpDjT0__camera_view_direction_change__00", "video_id": "uvo__2FGgpJpDjT0", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/a7c3763ac7fb87973e85/F0.jpg", "media/a7c3763ac7fb87973e85/F1.jpg", "media/a7c3763ac7fb87973e85/F2.jpg", "media/a7c3763ac7fb87973e85/F3.jpg", "media/a7c3763ac7fb87973e85/F4.jpg", "media/a7c3763ac7fb87973e85/F5.jpg", "media/a7c3763ac7fb87973e85/F6.jpg", "media/a7c3763ac7fb87973e85/F7.jpg", "media/a7c3763ac7fb87973e85/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "81da157a1e56b87762e7814ebb51c1b483860d4fc43dfa880d80ce362414b87b"}
{"qa_id": "uvo__2FMF6BhiVqs__object_world_motion_direction__00", "video_id": "uvo__2FMF6BhiVqs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/f673c6b1ccf3b707b36a/F0.jpg", "media/f673c6b1ccf3b707b36a/F1.jpg", "media/f673c6b1ccf3b707b36a/F2.jpg", "media/f673c6b1ccf3b707b36a/F3.jpg", "media/f673c6b1ccf3b707b36a/F4.jpg", "media/f673c6b1ccf3b707b36a/F5.jpg", "media/f673c6b1ccf3b707b36a/F6.jpg", "media/f673c6b1ccf3b707b36a/F7.jpg", "media/f673c6b1ccf3b707b36a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "16f42956c9d653e5ce380f300c66e68d6990639e463e100dfe7dc83513a92ee7"}
{"qa_id": "uvo__2FMF6BhiVqs__camera_object_distance_change__01", "video_id": "uvo__2FMF6BhiVqs", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (person in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/f673c6b1ccf3b707b36a/F0.jpg", "media/f673c6b1ccf3b707b36a/F1.jpg", "media/f673c6b1ccf3b707b36a/F2.jpg", "media/f673c6b1ccf3b707b36a/F3.jpg", "media/f673c6b1ccf3b707b36a/F4.jpg", "media/f673c6b1ccf3b707b36a/F5.jpg", "media/f673c6b1ccf3b707b36a/F6.jpg", "media/f673c6b1ccf3b707b36a/F7.jpg", "media/f673c6b1ccf3b707b36a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a86acb176c85d50230dd52345fa96485b4bb717e2fe4e73f66905e647c46119c"}
{"qa_id": "uvo__2HSFfjEXKlU__camera_translation_direction__00", "video_id": "uvo__2HSFfjEXKlU", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "up", "down"], "frame_paths": ["media/f6459de302fba36ebae0/F0.jpg", "media/f6459de302fba36ebae0/F1.jpg", "media/f6459de302fba36ebae0/F2.jpg", "media/f6459de302fba36ebae0/F3.jpg", "media/f6459de302fba36ebae0/F4.jpg", "media/f6459de302fba36ebae0/F5.jpg", "media/f6459de302fba36ebae0/F6.jpg", "media/f6459de302fba36ebae0/F7.jpg", "media/f6459de302fba36ebae0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "42d495577af686fd52ad42be790aca5b4c65d13eeff05998ce2780774a860623"}
{"qa_id": "uvo__2IjLUjatXNY__camera_view_direction_change__00", "video_id": "uvo__2IjLUjatXNY", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/39ee988a70170274376b/F0.jpg", "media/39ee988a70170274376b/F1.jpg", "media/39ee988a70170274376b/F2.jpg", "media/39ee988a70170274376b/F3.jpg", "media/39ee988a70170274376b/F4.jpg", "media/39ee988a70170274376b/F5.jpg", "media/39ee988a70170274376b/F6.jpg", "media/39ee988a70170274376b/F7.jpg", "media/39ee988a70170274376b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cb4db86f6ad842455c849665523a4999ea4c48a30c8629c691fd80577c13536b"}
{"qa_id": "uvo__2NeC5vzrP20__camera_translation_direction__00", "video_id": "uvo__2NeC5vzrP20", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "right", "forward", "left"], "frame_paths": ["media/f34eff0388523c4a6fa5/F0.jpg", "media/f34eff0388523c4a6fa5/F1.jpg", "media/f34eff0388523c4a6fa5/F2.jpg", "media/f34eff0388523c4a6fa5/F3.jpg", "media/f34eff0388523c4a6fa5/F4.jpg", "media/f34eff0388523c4a6fa5/F5.jpg", "media/f34eff0388523c4a6fa5/F6.jpg", "media/f34eff0388523c4a6fa5/F7.jpg", "media/f34eff0388523c4a6fa5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "086b0590fcdf23510fe882bb8fbfe840657e96dbb235f3576ea9022182ca9d65"}
{"qa_id": "uvo__2QeFt-2QcQw__camera_view_direction_change__00", "video_id": "uvo__2QeFt-2QcQw", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/8770c4b7eca3b581db2e/F0.jpg", "media/8770c4b7eca3b581db2e/F1.jpg", "media/8770c4b7eca3b581db2e/F2.jpg", "media/8770c4b7eca3b581db2e/F3.jpg", "media/8770c4b7eca3b581db2e/F4.jpg", "media/8770c4b7eca3b581db2e/F5.jpg", "media/8770c4b7eca3b581db2e/F6.jpg", "media/8770c4b7eca3b581db2e/F7.jpg", "media/8770c4b7eca3b581db2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "199105120d5aae11924ed2a33bf9bb468e8c513c19f24ed56588c61b3bf1f97e"}
{"qa_id": "uvo__2ZeGylHqo7M__object_world_motion_direction__00", "video_id": "uvo__2ZeGylHqo7M", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (person in red shirt and white cap) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "down", "up", "backward"], "frame_paths": ["media/286b6a8d66da7aa21f3a/F0.jpg", "media/286b6a8d66da7aa21f3a/F1.jpg", "media/286b6a8d66da7aa21f3a/F2.jpg", "media/286b6a8d66da7aa21f3a/F3.jpg", "media/286b6a8d66da7aa21f3a/F4.jpg", "media/286b6a8d66da7aa21f3a/F5.jpg", "media/286b6a8d66da7aa21f3a/F6.jpg", "media/286b6a8d66da7aa21f3a/F7.jpg", "media/286b6a8d66da7aa21f3a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (person in red shirt and white cap) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6b900dfecfa1a61b712bc1751b68645d4fd75ce4173761063456e8de580e1e43"}
{"qa_id": "uvo__2ZeGylHqo7M__camera_object_distance_change__01", "video_id": "uvo__2ZeGylHqo7M", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object B (person in red shirt and white cap) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/286b6a8d66da7aa21f3a/F0.jpg", "media/286b6a8d66da7aa21f3a/F1.jpg", "media/286b6a8d66da7aa21f3a/F2.jpg", "media/286b6a8d66da7aa21f3a/F3.jpg", "media/286b6a8d66da7aa21f3a/F4.jpg", "media/286b6a8d66da7aa21f3a/F5.jpg", "media/286b6a8d66da7aa21f3a/F6.jpg", "media/286b6a8d66da7aa21f3a/F7.jpg", "media/286b6a8d66da7aa21f3a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object B (person in red shirt and white cap) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "204801e5d706f73ea20c86d6fb66917b045b54fc8daee96668708e7a973ba629"}
{"qa_id": "uvo__2ZeGylHqo7M__object_pair_distance_change__02", "video_id": "uvo__2ZeGylHqo7M", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did the 3D distance between object A (person in red shirt and red cap) and object B (person in red shirt and white cap) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/1a54dff908ae2994baf1/F0.jpg", "media/1a54dff908ae2994baf1/F1.jpg", "media/1a54dff908ae2994baf1/F2.jpg", "media/1a54dff908ae2994baf1/F3.jpg", "media/1a54dff908ae2994baf1/F4.jpg", "media/1a54dff908ae2994baf1/F5.jpg", "media/1a54dff908ae2994baf1/F6.jpg", "media/1a54dff908ae2994baf1/F7.jpg", "media/1a54dff908ae2994baf1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did the 3D distance between object A (person in red shirt and red cap) and object B (person in red shirt and white cap) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5bd5362cfea0974097dd3a738c44aee2c24b67d47477eb39b007a9e0cedb8f00"}
{"qa_id": "uvo__2ZeGylHqo7M__object_travel_distance_comparison__03", "video_id": "uvo__2ZeGylHqo7M", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F5 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt and red cap) or B (person in red shirt and white cap)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1a54dff908ae2994baf1/F0.jpg", "media/1a54dff908ae2994baf1/F1.jpg", "media/1a54dff908ae2994baf1/F2.jpg", "media/1a54dff908ae2994baf1/F3.jpg", "media/1a54dff908ae2994baf1/F4.jpg", "media/1a54dff908ae2994baf1/F5.jpg", "media/1a54dff908ae2994baf1/F6.jpg", "media/1a54dff908ae2994baf1/F7.jpg", "media/1a54dff908ae2994baf1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in red shirt and red cap) or B (person in red shirt and white cap)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "50ee315afcebd32c598af15ad71acf8bfb84b81dbed10c1e014e8ea0e8b9625e"}
{"qa_id": "uvo__2bkfXxPwrLw__camera_view_direction_change__00", "video_id": "uvo__2bkfXxPwrLw", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/1ae681ee43f02bfbe128/F0.jpg", "media/1ae681ee43f02bfbe128/F1.jpg", "media/1ae681ee43f02bfbe128/F2.jpg", "media/1ae681ee43f02bfbe128/F3.jpg", "media/1ae681ee43f02bfbe128/F4.jpg", "media/1ae681ee43f02bfbe128/F5.jpg", "media/1ae681ee43f02bfbe128/F6.jpg", "media/1ae681ee43f02bfbe128/F7.jpg", "media/1ae681ee43f02bfbe128/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c57491379e0456fc779cb0a793c1f88519bfc10979e7513ddc4e99ad6d02821"}
{"qa_id": "uvo__2dFHJsTJbXM__camera_view_direction_change__00", "video_id": "uvo__2dFHJsTJbXM", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/7c91ab1de86523a01ed8/F0.jpg", "media/7c91ab1de86523a01ed8/F1.jpg", "media/7c91ab1de86523a01ed8/F2.jpg", "media/7c91ab1de86523a01ed8/F3.jpg", "media/7c91ab1de86523a01ed8/F4.jpg", "media/7c91ab1de86523a01ed8/F5.jpg", "media/7c91ab1de86523a01ed8/F6.jpg", "media/7c91ab1de86523a01ed8/F7.jpg", "media/7c91ab1de86523a01ed8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c84723bdba202b4de39dce825d2ddba78067da636f61db40c4ffb60077716de"}
{"qa_id": "uvo__2eBqLh-rrXU__object_world_motion_direction__00", "video_id": "uvo__2eBqLh-rrXU", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (small snowmobile with orange body) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "right", "left", "down"], "frame_paths": ["media/2173e92240fd5f070870/F0.jpg", "media/2173e92240fd5f070870/F1.jpg", "media/2173e92240fd5f070870/F2.jpg", "media/2173e92240fd5f070870/F3.jpg", "media/2173e92240fd5f070870/F4.jpg", "media/2173e92240fd5f070870/F5.jpg", "media/2173e92240fd5f070870/F6.jpg", "media/2173e92240fd5f070870/F7.jpg", "media/2173e92240fd5f070870/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (small snowmobile with orange body) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c114a3cbc6c216cd1954aebd4593ba46eec5053c97ac2fd944218eae9f181102"}
{"qa_id": "uvo__2eBqLh-rrXU__object_world_motion_direction__01", "video_id": "uvo__2eBqLh-rrXU", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (person wearing a green jacket riding the snowmobile) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "right", "down", "backward", "left", "forward"], "frame_paths": ["media/7a14fe1a5e60ae7b9eca/F0.jpg", "media/7a14fe1a5e60ae7b9eca/F1.jpg", "media/7a14fe1a5e60ae7b9eca/F2.jpg", "media/7a14fe1a5e60ae7b9eca/F3.jpg", "media/7a14fe1a5e60ae7b9eca/F4.jpg", "media/7a14fe1a5e60ae7b9eca/F5.jpg", "media/7a14fe1a5e60ae7b9eca/F6.jpg", "media/7a14fe1a5e60ae7b9eca/F7.jpg", "media/7a14fe1a5e60ae7b9eca/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (person wearing a green jacket riding the snowmobile) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. down\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "98115ea0384d29e321380c7a3f8c3720b9279815413f2ed54c8fe10e058af13d"}
{"qa_id": "uvo__2elua24dIC0__camera_view_direction_change__00", "video_id": "uvo__2elua24dIC0", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/dc34c4c027dcfb04c1c1/F0.jpg", "media/dc34c4c027dcfb04c1c1/F1.jpg", "media/dc34c4c027dcfb04c1c1/F2.jpg", "media/dc34c4c027dcfb04c1c1/F3.jpg", "media/dc34c4c027dcfb04c1c1/F4.jpg", "media/dc34c4c027dcfb04c1c1/F5.jpg", "media/dc34c4c027dcfb04c1c1/F6.jpg", "media/dc34c4c027dcfb04c1c1/F7.jpg", "media/dc34c4c027dcfb04c1c1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b5422fe5c1dbbc4213778859e6655aefc94a5b034daad0d8e5c6e93594695b9e"}
{"qa_id": "uvo__2fRD4L_7kOM__camera_view_direction_change__00", "video_id": "uvo__2fRD4L_7kOM", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/0338ccd47ecce0ddc787/F0.jpg", "media/0338ccd47ecce0ddc787/F1.jpg", "media/0338ccd47ecce0ddc787/F2.jpg", "media/0338ccd47ecce0ddc787/F3.jpg", "media/0338ccd47ecce0ddc787/F4.jpg", "media/0338ccd47ecce0ddc787/F5.jpg", "media/0338ccd47ecce0ddc787/F6.jpg", "media/0338ccd47ecce0ddc787/F7.jpg", "media/0338ccd47ecce0ddc787/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e075abf80b8f568c63df87a8c22d2a074ed86e6c7284c32de692757af209bfbd"}
{"qa_id": "uvo__2gDXSXMQOwo__camera_translation_direction__00", "video_id": "uvo__2gDXSXMQOwo", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "down", "up", "forward", "left", "backward"], "frame_paths": ["media/d812c7ed69fa89bf50de/F0.jpg", "media/d812c7ed69fa89bf50de/F1.jpg", "media/d812c7ed69fa89bf50de/F2.jpg", "media/d812c7ed69fa89bf50de/F3.jpg", "media/d812c7ed69fa89bf50de/F4.jpg", "media/d812c7ed69fa89bf50de/F5.jpg", "media/d812c7ed69fa89bf50de/F6.jpg", "media/d812c7ed69fa89bf50de/F7.jpg", "media/d812c7ed69fa89bf50de/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. up\nD. forward\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1886c84b85add1f9a7863dbf0b9a33362c124e001f7bb824275afdd00f07384"}
{"qa_id": "uvo__2gDXSXMQOwo__camera_view_direction_change__01", "video_id": "uvo__2gDXSXMQOwo", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/d812c7ed69fa89bf50de/F0.jpg", "media/d812c7ed69fa89bf50de/F1.jpg", "media/d812c7ed69fa89bf50de/F2.jpg", "media/d812c7ed69fa89bf50de/F3.jpg", "media/d812c7ed69fa89bf50de/F4.jpg", "media/d812c7ed69fa89bf50de/F5.jpg", "media/d812c7ed69fa89bf50de/F6.jpg", "media/d812c7ed69fa89bf50de/F7.jpg", "media/d812c7ed69fa89bf50de/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8357453a0a5b05085be349dfe49b047b1deec5007247482eaa1f6878a5bed78d"}
{"qa_id": "uvo__2kv48vBWTqs__object_world_motion_direction__00", "video_id": "uvo__2kv48vBWTqs", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object B (woman in black top and patterned skirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "left", "forward", "down"], "frame_paths": ["media/aea560e3ef9899de571c/F0.jpg", "media/aea560e3ef9899de571c/F1.jpg", "media/aea560e3ef9899de571c/F2.jpg", "media/aea560e3ef9899de571c/F3.jpg", "media/aea560e3ef9899de571c/F4.jpg", "media/aea560e3ef9899de571c/F5.jpg", "media/aea560e3ef9899de571c/F6.jpg", "media/aea560e3ef9899de571c/F7.jpg", "media/aea560e3ef9899de571c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object B (woman in black top and patterned skirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "10a8c74555bb079287c08b6f8be17f28b7d4f874b2d98d3ff413fb0575447d33"}
{"qa_id": "uvo__2kv48vBWTqs__object_travel_distance_comparison__01", "video_id": "uvo__2kv48vBWTqs", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in white t-shirt and dark pants) or B (woman in black top and patterned skirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/68326ae38e8d5815e072/F0.jpg", "media/68326ae38e8d5815e072/F1.jpg", "media/68326ae38e8d5815e072/F2.jpg", "media/68326ae38e8d5815e072/F3.jpg", "media/68326ae38e8d5815e072/F4.jpg", "media/68326ae38e8d5815e072/F5.jpg", "media/68326ae38e8d5815e072/F6.jpg", "media/68326ae38e8d5815e072/F7.jpg", "media/68326ae38e8d5815e072/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in white t-shirt and dark pants) or B (woman in black top and patterned skirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "61e5c6f10522615c7b53734b899e5ed9ef32d348f8b7749042c5aa1d4db260e6"}
{"qa_id": "uvo__2tNj-ludBKQ__camera_translation_direction__00", "video_id": "uvo__2tNj-ludBKQ", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "left", "up", "backward"], "frame_paths": ["media/b83280ee48a5ec2095a5/F0.jpg", "media/b83280ee48a5ec2095a5/F1.jpg", "media/b83280ee48a5ec2095a5/F2.jpg", "media/b83280ee48a5ec2095a5/F3.jpg", "media/b83280ee48a5ec2095a5/F4.jpg", "media/b83280ee48a5ec2095a5/F5.jpg", "media/b83280ee48a5ec2095a5/F6.jpg", "media/b83280ee48a5ec2095a5/F7.jpg", "media/b83280ee48a5ec2095a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "34160dea160fd38f1f1f8f0573017054bf9fb5a5a32c5e64eb7ecc0658cc0882"}
{"qa_id": "uvo__2uYG51S03kI__object_world_motion_direction__00", "video_id": "uvo__2uYG51S03kI", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (large orange dump truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "backward", "down", "left"], "frame_paths": ["media/35ffffcf303f5338f499/F0.jpg", "media/35ffffcf303f5338f499/F1.jpg", "media/35ffffcf303f5338f499/F2.jpg", "media/35ffffcf303f5338f499/F3.jpg", "media/35ffffcf303f5338f499/F4.jpg", "media/35ffffcf303f5338f499/F5.jpg", "media/35ffffcf303f5338f499/F6.jpg", "media/35ffffcf303f5338f499/F7.jpg", "media/35ffffcf303f5338f499/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (large orange dump truck) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88e5280d4ad95f9183e0ea82f5ca7090773dfc5f8fa2908b2bbfff3cdebdf0b8"}
{"qa_id": "uvo__2vJgJiJkxlI__camera_view_direction_change__00", "video_id": "uvo__2vJgJiJkxlI", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "left", "down", "right"], "frame_paths": ["media/44624e44fb5956390e84/F0.jpg", "media/44624e44fb5956390e84/F1.jpg", "media/44624e44fb5956390e84/F2.jpg", "media/44624e44fb5956390e84/F3.jpg", "media/44624e44fb5956390e84/F4.jpg", "media/44624e44fb5956390e84/F5.jpg", "media/44624e44fb5956390e84/F6.jpg", "media/44624e44fb5956390e84/F7.jpg", "media/44624e44fb5956390e84/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. left\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38b818f9dcd72d220ac2973026183b7423b584460f12c67f5931aac1d0058b2f"}
{"qa_id": "uvo__2wGS8QoKkMo__camera_view_direction_change__00", "video_id": "uvo__2wGS8QoKkMo", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/0117dec9fe201c9b35c2/F0.jpg", "media/0117dec9fe201c9b35c2/F1.jpg", "media/0117dec9fe201c9b35c2/F2.jpg", "media/0117dec9fe201c9b35c2/F3.jpg", "media/0117dec9fe201c9b35c2/F4.jpg", "media/0117dec9fe201c9b35c2/F5.jpg", "media/0117dec9fe201c9b35c2/F6.jpg", "media/0117dec9fe201c9b35c2/F7.jpg", "media/0117dec9fe201c9b35c2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__30tjJR1Az8o__camera_view_direction_change__00", "video_id": "uvo__30tjJR1Az8o", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/83d228393bb9adacc368/F0.jpg", "media/83d228393bb9adacc368/F1.jpg", "media/83d228393bb9adacc368/F2.jpg", "media/83d228393bb9adacc368/F3.jpg", "media/83d228393bb9adacc368/F4.jpg", "media/83d228393bb9adacc368/F5.jpg", "media/83d228393bb9adacc368/F6.jpg", "media/83d228393bb9adacc368/F7.jpg", "media/83d228393bb9adacc368/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "654016ae105affb53bc2face594c67e96f587f4cdc806a6f418076b30032d255"}
{"qa_id": "uvo__32X8SJ-nFGI__camera_translation_direction__00", "video_id": "uvo__32X8SJ-nFGI", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "left", "backward", "down", "up", "forward"], "frame_paths": ["media/332635ed6678118bf204/F0.jpg", "media/332635ed6678118bf204/F1.jpg", "media/332635ed6678118bf204/F2.jpg", "media/332635ed6678118bf204/F3.jpg", "media/332635ed6678118bf204/F4.jpg", "media/332635ed6678118bf204/F5.jpg", "media/332635ed6678118bf204/F6.jpg", "media/332635ed6678118bf204/F7.jpg", "media/332635ed6678118bf204/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. backward\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cd83f625e58ad3a0f23a3d968051bae4f8c609d9bd787db3576039ddc390e550"}
{"qa_id": "uvo__35QqEr2eIXQ__camera_translation_direction__00", "video_id": "uvo__35QqEr2eIXQ", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "forward", "right", "left"], "frame_paths": ["media/12ecefacabb27ef28b75/F0.jpg", "media/12ecefacabb27ef28b75/F1.jpg", "media/12ecefacabb27ef28b75/F2.jpg", "media/12ecefacabb27ef28b75/F3.jpg", "media/12ecefacabb27ef28b75/F4.jpg", "media/12ecefacabb27ef28b75/F5.jpg", "media/12ecefacabb27ef28b75/F6.jpg", "media/12ecefacabb27ef28b75/F7.jpg", "media/12ecefacabb27ef28b75/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78b747b66fa1d9a31c59df929ee1a120cb7dc10b6992a23ba4cc8c4ff4bb8338"}
{"qa_id": "uvo__39TjWkmBseY__camera_view_direction_change__00", "video_id": "uvo__39TjWkmBseY", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/6449cfc4a8d0f9e36fa0/F0.jpg", "media/6449cfc4a8d0f9e36fa0/F1.jpg", "media/6449cfc4a8d0f9e36fa0/F2.jpg", "media/6449cfc4a8d0f9e36fa0/F3.jpg", "media/6449cfc4a8d0f9e36fa0/F4.jpg", "media/6449cfc4a8d0f9e36fa0/F5.jpg", "media/6449cfc4a8d0f9e36fa0/F6.jpg", "media/6449cfc4a8d0f9e36fa0/F7.jpg", "media/6449cfc4a8d0f9e36fa0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "628412a7a6a40e3160b77784c8f4351610f72858cb6929cf596ee4afe5c2ce91"}
{"qa_id": "uvo__39TjWkmBseY__object_world_motion_direction__01", "video_id": "uvo__39TjWkmBseY", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (person wearing an orange jacket riding a snowmobile) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "right", "up", "down", "left"], "frame_paths": ["media/5f14db312b089b40ce37/F0.jpg", "media/5f14db312b089b40ce37/F1.jpg", "media/5f14db312b089b40ce37/F2.jpg", "media/5f14db312b089b40ce37/F3.jpg", "media/5f14db312b089b40ce37/F4.jpg", "media/5f14db312b089b40ce37/F5.jpg", "media/5f14db312b089b40ce37/F6.jpg", "media/5f14db312b089b40ce37/F7.jpg", "media/5f14db312b089b40ce37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (person wearing an orange jacket riding a snowmobile) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. right\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d6371f8dc3c2e3340e5520b06e6fa372bb3dcfbde3f213a7e0a84eebdae4cbe"}
{"qa_id": "uvo__39TjWkmBseY__camera_object_distance_change__02", "video_id": "uvo__39TjWkmBseY", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did object A (person wearing an orange jacket riding a snowmobile) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/5f14db312b089b40ce37/F0.jpg", "media/5f14db312b089b40ce37/F1.jpg", "media/5f14db312b089b40ce37/F2.jpg", "media/5f14db312b089b40ce37/F3.jpg", "media/5f14db312b089b40ce37/F4.jpg", "media/5f14db312b089b40ce37/F5.jpg", "media/5f14db312b089b40ce37/F6.jpg", "media/5f14db312b089b40ce37/F7.jpg", "media/5f14db312b089b40ce37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (person wearing an orange jacket riding a snowmobile) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "30b9a185851c8541234f474b4227e128ecd10b3deef6ccb93c8e95da42473429"}
{"qa_id": "uvo__39TjWkmBseY__object_world_motion_direction__03", "video_id": "uvo__39TjWkmBseY", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (black snowmobile being driven on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "left", "forward", "right"], "frame_paths": ["media/99f038843637b19768e4/F0.jpg", "media/99f038843637b19768e4/F1.jpg", "media/99f038843637b19768e4/F2.jpg", "media/99f038843637b19768e4/F3.jpg", "media/99f038843637b19768e4/F4.jpg", "media/99f038843637b19768e4/F5.jpg", "media/99f038843637b19768e4/F6.jpg", "media/99f038843637b19768e4/F7.jpg", "media/99f038843637b19768e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (black snowmobile being driven on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. left\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7b14616f1db324086d81b9f2a90b06529329688a9167273481c29d3b26b8e956"}
{"qa_id": "uvo__39TjWkmBseY__camera_object_distance_change__04", "video_id": "uvo__39TjWkmBseY", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did object B (black snowmobile being driven on snow) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/99f038843637b19768e4/F0.jpg", "media/99f038843637b19768e4/F1.jpg", "media/99f038843637b19768e4/F2.jpg", "media/99f038843637b19768e4/F3.jpg", "media/99f038843637b19768e4/F4.jpg", "media/99f038843637b19768e4/F5.jpg", "media/99f038843637b19768e4/F6.jpg", "media/99f038843637b19768e4/F7.jpg", "media/99f038843637b19768e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object B (black snowmobile being driven on snow) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "98ad909f29cd46c28d6ee212f046fc3b82bcf2c21c39aba3eddb56768eb23604"}
{"qa_id": "uvo__3ASMnwf7lo8__camera_translation_direction__00", "video_id": "uvo__3ASMnwf7lo8", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "up", "backward", "right"], "frame_paths": ["media/84a9498e85a60f4527c9/F0.jpg", "media/84a9498e85a60f4527c9/F1.jpg", "media/84a9498e85a60f4527c9/F2.jpg", "media/84a9498e85a60f4527c9/F3.jpg", "media/84a9498e85a60f4527c9/F4.jpg", "media/84a9498e85a60f4527c9/F5.jpg", "media/84a9498e85a60f4527c9/F6.jpg", "media/84a9498e85a60f4527c9/F7.jpg", "media/84a9498e85a60f4527c9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. up\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "68f1ac42668e9b92932e4849c45d2560cccb93a6d0db8340ce64858d36df334d"}
{"qa_id": "uvo__3ASMnwf7lo8__camera_object_distance_change__01", "video_id": "uvo__3ASMnwf7lo8", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did object A (woman in a dark blue tennis dress holding a racket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9c1013134536700c9836/F0.jpg", "media/9c1013134536700c9836/F1.jpg", "media/9c1013134536700c9836/F2.jpg", "media/9c1013134536700c9836/F3.jpg", "media/9c1013134536700c9836/F4.jpg", "media/9c1013134536700c9836/F5.jpg", "media/9c1013134536700c9836/F6.jpg", "media/9c1013134536700c9836/F7.jpg", "media/9c1013134536700c9836/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (woman in a dark blue tennis dress holding a racket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e6f34b3b0a73a83b0998a1b85224f9fb58ee2fa9c6ee138a4fd3aebccd65c9f"}
{"qa_id": "uvo__3ASMnwf7lo8__object_pair_distance_change__02", "video_id": "uvo__3ASMnwf7lo8", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F5, did the 3D distance between object A (woman in a dark blue tennis dress holding a racket) and object B (woman in a bright green polo shirt and white pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/06cbf55348c1cbcc230c/F0.jpg", "media/06cbf55348c1cbcc230c/F1.jpg", "media/06cbf55348c1cbcc230c/F2.jpg", "media/06cbf55348c1cbcc230c/F3.jpg", "media/06cbf55348c1cbcc230c/F4.jpg", "media/06cbf55348c1cbcc230c/F5.jpg", "media/06cbf55348c1cbcc230c/F6.jpg", "media/06cbf55348c1cbcc230c/F7.jpg", "media/06cbf55348c1cbcc230c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did the 3D distance between object A (woman in a dark blue tennis dress holding a racket) and object B (woman in a bright green polo shirt and white pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "29f9481c21269141c24e1164d4f236debb22b78f09efd4ca27f45f64820acbeb"}
{"qa_id": "uvo__3GTEJEh-bC8__camera_translation_direction__00", "video_id": "uvo__3GTEJEh-bC8", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "left", "down", "backward"], "frame_paths": ["media/9147360fd705b00ba624/F0.jpg", "media/9147360fd705b00ba624/F1.jpg", "media/9147360fd705b00ba624/F2.jpg", "media/9147360fd705b00ba624/F3.jpg", "media/9147360fd705b00ba624/F4.jpg", "media/9147360fd705b00ba624/F5.jpg", "media/9147360fd705b00ba624/F6.jpg", "media/9147360fd705b00ba624/F7.jpg", "media/9147360fd705b00ba624/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bdbc303c0b8fea5f881007b4e014b0e6e87c462ed8410077ad4cb5902303c560"}
{"qa_id": "uvo__3GTEJEh-bC8__object_world_motion_direction__01", "video_id": "uvo__3GTEJEh-bC8", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in dark clothing with a bright headlamp) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "forward", "backward", "up"], "frame_paths": ["media/c2fea7238cf7e367cf7c/F0.jpg", "media/c2fea7238cf7e367cf7c/F1.jpg", "media/c2fea7238cf7e367cf7c/F2.jpg", "media/c2fea7238cf7e367cf7c/F3.jpg", "media/c2fea7238cf7e367cf7c/F4.jpg", "media/c2fea7238cf7e367cf7c/F5.jpg", "media/c2fea7238cf7e367cf7c/F6.jpg", "media/c2fea7238cf7e367cf7c/F7.jpg", "media/c2fea7238cf7e367cf7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (person in dark clothing with a bright headlamp) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "72f64bc4c72c86678d9b65a21d1c1328dc9bdcadc2541304c8acad61583e1044"}
{"qa_id": "uvo__3GTEJEh-bC8__camera_object_distance_change__02", "video_id": "uvo__3GTEJEh-bC8", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (person in dark clothing with a bright headlamp) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c2fea7238cf7e367cf7c/F0.jpg", "media/c2fea7238cf7e367cf7c/F1.jpg", "media/c2fea7238cf7e367cf7c/F2.jpg", "media/c2fea7238cf7e367cf7c/F3.jpg", "media/c2fea7238cf7e367cf7c/F4.jpg", "media/c2fea7238cf7e367cf7c/F5.jpg", "media/c2fea7238cf7e367cf7c/F6.jpg", "media/c2fea7238cf7e367cf7c/F7.jpg", "media/c2fea7238cf7e367cf7c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (person in dark clothing with a bright headlamp) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c6b98b4b2de4a678f0196a1170214961e94ca09f1dca11af80b59ed77e94da87"}
{"qa_id": "uvo__3k8PCuYldBs__camera_translation_direction__00", "video_id": "uvo__3k8PCuYldBs", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "backward", "right", "left", "down"], "frame_paths": ["media/0db59528f586c054b957/F0.jpg", "media/0db59528f586c054b957/F1.jpg", "media/0db59528f586c054b957/F2.jpg", "media/0db59528f586c054b957/F3.jpg", "media/0db59528f586c054b957/F4.jpg", "media/0db59528f586c054b957/F5.jpg", "media/0db59528f586c054b957/F6.jpg", "media/0db59528f586c054b957/F7.jpg", "media/0db59528f586c054b957/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. backward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ae85e0824e0a68b48695a91ca231f859b1752b530bbb7a5eca3e8e002b252404"}
{"qa_id": "uvo__3k8PCuYldBs__camera_view_direction_change__01", "video_id": "uvo__3k8PCuYldBs", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/0db59528f586c054b957/F0.jpg", "media/0db59528f586c054b957/F1.jpg", "media/0db59528f586c054b957/F2.jpg", "media/0db59528f586c054b957/F3.jpg", "media/0db59528f586c054b957/F4.jpg", "media/0db59528f586c054b957/F5.jpg", "media/0db59528f586c054b957/F6.jpg", "media/0db59528f586c054b957/F7.jpg", "media/0db59528f586c054b957/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9fca55b1611ae8dd2346bf922298894d229a7f1fe48f897ccdbc4dcbbbea48d9"}
{"qa_id": "uvo__3k8PCuYldBs__object_world_motion_direction__02", "video_id": "uvo__3k8PCuYldBs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (horse with a rider in a blue shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "backward", "forward", "up"], "frame_paths": ["media/4a761d76e8f9cc7d89e6/F0.jpg", "media/4a761d76e8f9cc7d89e6/F1.jpg", "media/4a761d76e8f9cc7d89e6/F2.jpg", "media/4a761d76e8f9cc7d89e6/F3.jpg", "media/4a761d76e8f9cc7d89e6/F4.jpg", "media/4a761d76e8f9cc7d89e6/F5.jpg", "media/4a761d76e8f9cc7d89e6/F6.jpg", "media/4a761d76e8f9cc7d89e6/F7.jpg", "media/4a761d76e8f9cc7d89e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (horse with a rider in a blue shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d907470a2dd9c1cc3249e829680af4b6e0712c4f91f2797e48a5e0c509c4feb"}
{"qa_id": "uvo__3k8PCuYldBs__object_world_motion_direction__03", "video_id": "uvo__3k8PCuYldBs", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object B (horse in the foreground, viewed from behind) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "up", "down", "left"], "frame_paths": ["media/8cd98b4f8165a634cf9d/F0.jpg", "media/8cd98b4f8165a634cf9d/F1.jpg", "media/8cd98b4f8165a634cf9d/F2.jpg", "media/8cd98b4f8165a634cf9d/F3.jpg", "media/8cd98b4f8165a634cf9d/F4.jpg", "media/8cd98b4f8165a634cf9d/F5.jpg", "media/8cd98b4f8165a634cf9d/F6.jpg", "media/8cd98b4f8165a634cf9d/F7.jpg", "media/8cd98b4f8165a634cf9d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object B (horse in the foreground, viewed from behind) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e37e7a698597a66995b4830f53159eb5f66dd845f6b4dbebec2c601898a3ac75"}
{"qa_id": "uvo__3kLCncjQYNg__camera_view_direction_change__00", "video_id": "uvo__3kLCncjQYNg", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/ade2a26cc1e06bf368fa/F0.jpg", "media/ade2a26cc1e06bf368fa/F1.jpg", "media/ade2a26cc1e06bf368fa/F2.jpg", "media/ade2a26cc1e06bf368fa/F3.jpg", "media/ade2a26cc1e06bf368fa/F4.jpg", "media/ade2a26cc1e06bf368fa/F5.jpg", "media/ade2a26cc1e06bf368fa/F6.jpg", "media/ade2a26cc1e06bf368fa/F7.jpg", "media/ade2a26cc1e06bf368fa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74e7834987ad2bdda409db81a3760e1ca41987794e8ca3e25f0c1204c3290113"}
{"qa_id": "uvo__3mZJgS61jVU__object_world_motion_direction__00", "video_id": "uvo__3mZJgS61jVU", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in pink jacket on orange sled) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "left", "right", "forward", "down", "backward"], "frame_paths": ["media/9b88213c679d8cdf56df/F0.jpg", "media/9b88213c679d8cdf56df/F1.jpg", "media/9b88213c679d8cdf56df/F2.jpg", "media/9b88213c679d8cdf56df/F3.jpg", "media/9b88213c679d8cdf56df/F4.jpg", "media/9b88213c679d8cdf56df/F5.jpg", "media/9b88213c679d8cdf56df/F6.jpg", "media/9b88213c679d8cdf56df/F7.jpg", "media/9b88213c679d8cdf56df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in pink jacket on orange sled) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. right\nD. forward\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c8170cf1e16427497a6d836516df0c18ceaa0c67fd190d1b553dc16db5c16476"}
{"qa_id": "uvo__3mZJgS61jVU__camera_object_distance_change__01", "video_id": "uvo__3mZJgS61jVU", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F3, did object A (person in pink jacket on orange sled) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9b88213c679d8cdf56df/F0.jpg", "media/9b88213c679d8cdf56df/F1.jpg", "media/9b88213c679d8cdf56df/F2.jpg", "media/9b88213c679d8cdf56df/F3.jpg", "media/9b88213c679d8cdf56df/F4.jpg", "media/9b88213c679d8cdf56df/F5.jpg", "media/9b88213c679d8cdf56df/F6.jpg", "media/9b88213c679d8cdf56df/F7.jpg", "media/9b88213c679d8cdf56df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in pink jacket on orange sled) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ef8dc6e4a25cb0b946bbe91d40b5f5427b6650aa64d25e9b6df493683a9e0caf"}
{"qa_id": "uvo__3oERY0un6Yo__object_world_motion_direction__00", "video_id": "uvo__3oERY0un6Yo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (small black dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "up", "forward"], "frame_paths": ["media/0409d2255f0bdec0f149/F0.jpg", "media/0409d2255f0bdec0f149/F1.jpg", "media/0409d2255f0bdec0f149/F2.jpg", "media/0409d2255f0bdec0f149/F3.jpg", "media/0409d2255f0bdec0f149/F4.jpg", "media/0409d2255f0bdec0f149/F5.jpg", "media/0409d2255f0bdec0f149/F6.jpg", "media/0409d2255f0bdec0f149/F7.jpg", "media/0409d2255f0bdec0f149/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (small black dog) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a1f250c2ab16e6c38c7bc6b7c92017a78f9e94bc0012b63ddd0b09b32143cedd"}
{"qa_id": "uvo__3oERY0un6Yo__object_world_motion_direction__01", "video_id": "uvo__3oERY0un6Yo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object B (person in light blue shirt and shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "down", "up", "left"], "frame_paths": ["media/3f61e25150fec76e8caa/F0.jpg", "media/3f61e25150fec76e8caa/F1.jpg", "media/3f61e25150fec76e8caa/F2.jpg", "media/3f61e25150fec76e8caa/F3.jpg", "media/3f61e25150fec76e8caa/F4.jpg", "media/3f61e25150fec76e8caa/F5.jpg", "media/3f61e25150fec76e8caa/F6.jpg", "media/3f61e25150fec76e8caa/F7.jpg", "media/3f61e25150fec76e8caa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object B (person in light blue shirt and shorts) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d92b0f45682e9f0e267c917474a8c635c7f64cfdd605964c6832198a29e92c73"}
{"qa_id": "uvo__3xR6gZ22Tdc__camera_view_direction_change__00", "video_id": "uvo__3xR6gZ22Tdc", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/e4dbd189d86be4257611/F0.jpg", "media/e4dbd189d86be4257611/F1.jpg", "media/e4dbd189d86be4257611/F2.jpg", "media/e4dbd189d86be4257611/F3.jpg", "media/e4dbd189d86be4257611/F4.jpg", "media/e4dbd189d86be4257611/F5.jpg", "media/e4dbd189d86be4257611/F6.jpg", "media/e4dbd189d86be4257611/F7.jpg", "media/e4dbd189d86be4257611/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88b40c1ef52643cefe4ef1c3fa249f1a470c7fcca2783a307d04d9b65d9496e2"}
{"qa_id": "uvo__40SoUIggUEw__camera_translation_direction__00", "video_id": "uvo__40SoUIggUEw", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "up", "forward"], "frame_paths": ["media/347e58743946fb926d45/F0.jpg", "media/347e58743946fb926d45/F1.jpg", "media/347e58743946fb926d45/F2.jpg", "media/347e58743946fb926d45/F3.jpg", "media/347e58743946fb926d45/F4.jpg", "media/347e58743946fb926d45/F5.jpg", "media/347e58743946fb926d45/F6.jpg", "media/347e58743946fb926d45/F7.jpg", "media/347e58743946fb926d45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eadf2470e4a6058f210ba57104b7ff9a1daa5cc6aff6cb6a67b2d603e2586633"}
{"qa_id": "uvo__40SoUIggUEw__object_world_motion_direction__01", "video_id": "uvo__40SoUIggUEw", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown horse with a white mane) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "down", "up", "left"], "frame_paths": ["media/7e3eb5b58965d9b226f5/F0.jpg", "media/7e3eb5b58965d9b226f5/F1.jpg", "media/7e3eb5b58965d9b226f5/F2.jpg", "media/7e3eb5b58965d9b226f5/F3.jpg", "media/7e3eb5b58965d9b226f5/F4.jpg", "media/7e3eb5b58965d9b226f5/F5.jpg", "media/7e3eb5b58965d9b226f5/F6.jpg", "media/7e3eb5b58965d9b226f5/F7.jpg", "media/7e3eb5b58965d9b226f5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (light brown horse with a white mane) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. down\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "75e174f8c2cb477834aee4e681a29c3e9e5bd9ffca96c653acf7b2d6da5b783c"}
{"qa_id": "uvo__40SoUIggUEw__object_world_motion_direction__02", "video_id": "uvo__40SoUIggUEw", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (dark brown horse ridden by a person in light clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "up", "backward"], "frame_paths": ["media/2e6756eaecd887dcaf6a/F0.jpg", "media/2e6756eaecd887dcaf6a/F1.jpg", "media/2e6756eaecd887dcaf6a/F2.jpg", "media/2e6756eaecd887dcaf6a/F3.jpg", "media/2e6756eaecd887dcaf6a/F4.jpg", "media/2e6756eaecd887dcaf6a/F5.jpg", "media/2e6756eaecd887dcaf6a/F6.jpg", "media/2e6756eaecd887dcaf6a/F7.jpg", "media/2e6756eaecd887dcaf6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (dark brown horse ridden by a person in light clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "014a15da173047a4a9c14455eb5b8b785f9b12abe351b7a4c69cc29f47311ad0"}
{"qa_id": "uvo__40SoUIggUEw__camera_object_distance_change__03", "video_id": "uvo__40SoUIggUEw", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object B (dark brown horse ridden by a person in light clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/2e6756eaecd887dcaf6a/F0.jpg", "media/2e6756eaecd887dcaf6a/F1.jpg", "media/2e6756eaecd887dcaf6a/F2.jpg", "media/2e6756eaecd887dcaf6a/F3.jpg", "media/2e6756eaecd887dcaf6a/F4.jpg", "media/2e6756eaecd887dcaf6a/F5.jpg", "media/2e6756eaecd887dcaf6a/F6.jpg", "media/2e6756eaecd887dcaf6a/F7.jpg", "media/2e6756eaecd887dcaf6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object B (dark brown horse ridden by a person in light clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6bb715e766d5228cf7835de0d628ea9478d7855cec8e5eed4c866a46198295fb"}
{"qa_id": "uvo__40SoUIggUEw__object_pair_distance_change__04", "video_id": "uvo__40SoUIggUEw", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did the 3D distance between object A (light brown horse with a white mane) and object B (dark brown horse ridden by a person in light clothing) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/bbbfe7fe91f65aaea127/F0.jpg", "media/bbbfe7fe91f65aaea127/F1.jpg", "media/bbbfe7fe91f65aaea127/F2.jpg", "media/bbbfe7fe91f65aaea127/F3.jpg", "media/bbbfe7fe91f65aaea127/F4.jpg", "media/bbbfe7fe91f65aaea127/F5.jpg", "media/bbbfe7fe91f65aaea127/F6.jpg", "media/bbbfe7fe91f65aaea127/F7.jpg", "media/bbbfe7fe91f65aaea127/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (light brown horse with a white mane) and object B (dark brown horse ridden by a person in light clothing) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ed070b9d552ac3a25a66c8b2db6ea53c4f1f1099daae4bf8ddb3d63a596c8274"}
{"qa_id": "uvo__40SoUIggUEw__object_travel_distance_comparison__05", "video_id": "uvo__40SoUIggUEw", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown horse with a white mane) or B (dark brown horse ridden by a person in light clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/bbbfe7fe91f65aaea127/F0.jpg", "media/bbbfe7fe91f65aaea127/F1.jpg", "media/bbbfe7fe91f65aaea127/F2.jpg", "media/bbbfe7fe91f65aaea127/F3.jpg", "media/bbbfe7fe91f65aaea127/F4.jpg", "media/bbbfe7fe91f65aaea127/F5.jpg", "media/bbbfe7fe91f65aaea127/F6.jpg", "media/bbbfe7fe91f65aaea127/F7.jpg", "media/bbbfe7fe91f65aaea127/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (light brown horse with a white mane) or B (dark brown horse ridden by a person in light clothing)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b8b9d4ee78f9417dcc96879efa62469266e48eda02fb2a11a503f862ad1b5cc"}
{"qa_id": "uvo__41YxghjkouA__camera_object_distance_change__00", "video_id": "uvo__41YxghjkouA", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object B (person in dark blue shirt and red shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/4a964dad07ebdad06d66/F0.jpg", "media/4a964dad07ebdad06d66/F1.jpg", "media/4a964dad07ebdad06d66/F2.jpg", "media/4a964dad07ebdad06d66/F3.jpg", "media/4a964dad07ebdad06d66/F4.jpg", "media/4a964dad07ebdad06d66/F5.jpg", "media/4a964dad07ebdad06d66/F6.jpg", "media/4a964dad07ebdad06d66/F7.jpg", "media/4a964dad07ebdad06d66/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object B (person in dark blue shirt and red shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3b6f13f5ced2b1ea691b1a49f43a252f14ef6cda9a64598db47e2352c67492b8"}
{"qa_id": "uvo__41uHZ_so8mc__object_pair_distance_change__00", "video_id": "uvo__41uHZ_so8mc", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did the 3D distance between object A (boy in blue rash guard and patterned swim trunks) and object B (boy with short dark hair in the water) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/595b0898c3b0a44d654e/F0.jpg", "media/595b0898c3b0a44d654e/F1.jpg", "media/595b0898c3b0a44d654e/F2.jpg", "media/595b0898c3b0a44d654e/F3.jpg", "media/595b0898c3b0a44d654e/F4.jpg", "media/595b0898c3b0a44d654e/F5.jpg", "media/595b0898c3b0a44d654e/F6.jpg", "media/595b0898c3b0a44d654e/F7.jpg", "media/595b0898c3b0a44d654e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g1", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (boy in blue rash guard and patterned swim trunks) and object B (boy with short dark hair in the water) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b1ff0b2e93644bc98e17786cb8bd530ee71791f41b49edb926f35bbfebc8c7bc"}
{"qa_id": "uvo__43QnKFB8tW8__object_world_motion_direction__00", "video_id": "uvo__43QnKFB8tW8", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (man with white hair wearing a black coat and blue tie) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "backward", "down", "right"], "frame_paths": ["media/2f01138a5a147b764c92/F0.jpg", "media/2f01138a5a147b764c92/F1.jpg", "media/2f01138a5a147b764c92/F2.jpg", "media/2f01138a5a147b764c92/F3.jpg", "media/2f01138a5a147b764c92/F4.jpg", "media/2f01138a5a147b764c92/F5.jpg", "media/2f01138a5a147b764c92/F6.jpg", "media/2f01138a5a147b764c92/F7.jpg", "media/2f01138a5a147b764c92/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (man with white hair wearing a black coat and blue tie) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. backward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc1fa92dac2e9a35f344756adae23183be48bc6b6f0c6ff87f14428bf8e7a19d"}
{"qa_id": "uvo__44qN_OJUGLk__camera_view_direction_change__00", "video_id": "uvo__44qN_OJUGLk", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/99a3cd43e29c474b65c9/F0.jpg", "media/99a3cd43e29c474b65c9/F1.jpg", "media/99a3cd43e29c474b65c9/F2.jpg", "media/99a3cd43e29c474b65c9/F3.jpg", "media/99a3cd43e29c474b65c9/F4.jpg", "media/99a3cd43e29c474b65c9/F5.jpg", "media/99a3cd43e29c474b65c9/F6.jpg", "media/99a3cd43e29c474b65c9/F7.jpg", "media/99a3cd43e29c474b65c9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa5bbe0f773427eec251354ffb864b848dea2a22cd4db093baa7aa2643832758"}
{"qa_id": "uvo__4OT7DEf8rCw__camera_view_direction_change__00", "video_id": "uvo__4OT7DEf8rCw", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/e74e169a90d39bcd426c/F0.jpg", "media/e74e169a90d39bcd426c/F1.jpg", "media/e74e169a90d39bcd426c/F2.jpg", "media/e74e169a90d39bcd426c/F3.jpg", "media/e74e169a90d39bcd426c/F4.jpg", "media/e74e169a90d39bcd426c/F5.jpg", "media/e74e169a90d39bcd426c/F6.jpg", "media/e74e169a90d39bcd426c/F7.jpg", "media/e74e169a90d39bcd426c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "uvo__4SVeKOmOclI__camera_view_direction_change__00", "video_id": "uvo__4SVeKOmOclI", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/2669eb536286e3e69867/F0.jpg", "media/2669eb536286e3e69867/F1.jpg", "media/2669eb536286e3e69867/F2.jpg", "media/2669eb536286e3e69867/F3.jpg", "media/2669eb536286e3e69867/F4.jpg", "media/2669eb536286e3e69867/F5.jpg", "media/2669eb536286e3e69867/F6.jpg", "media/2669eb536286e3e69867/F7.jpg", "media/2669eb536286e3e69867/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "139d78c609346a36b31a51baa6a9cfbf5b99d316434cbd3d4bd9ef34f83af385"}
{"qa_id": "uvo__4VZUCNBf1P4__camera_translation_direction__00", "video_id": "uvo__4VZUCNBf1P4", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "down", "up", "backward"], "frame_paths": ["media/9e5e71925fb094a6f3e1/F0.jpg", "media/9e5e71925fb094a6f3e1/F1.jpg", "media/9e5e71925fb094a6f3e1/F2.jpg", "media/9e5e71925fb094a6f3e1/F3.jpg", "media/9e5e71925fb094a6f3e1/F4.jpg", "media/9e5e71925fb094a6f3e1/F5.jpg", "media/9e5e71925fb094a6f3e1/F6.jpg", "media/9e5e71925fb094a6f3e1/F7.jpg", "media/9e5e71925fb094a6f3e1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. down\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0d585df46d852ae30ce8be8df47282f38a5fb66a8146b29b37d20f8cad9cfe7c"}
{"qa_id": "uvo__4VZUCNBf1P4__camera_view_direction_change__01", "video_id": "uvo__4VZUCNBf1P4", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/9e5e71925fb094a6f3e1/F0.jpg", "media/9e5e71925fb094a6f3e1/F1.jpg", "media/9e5e71925fb094a6f3e1/F2.jpg", "media/9e5e71925fb094a6f3e1/F3.jpg", "media/9e5e71925fb094a6f3e1/F4.jpg", "media/9e5e71925fb094a6f3e1/F5.jpg", "media/9e5e71925fb094a6f3e1/F6.jpg", "media/9e5e71925fb094a6f3e1/F7.jpg", "media/9e5e71925fb094a6f3e1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7f98011f6f5e3bef7856bffbee3a8c2579537b239284b48df76fd5bbadf4bc3f"}
{"qa_id": "uvo__4VZUCNBf1P4__object_world_motion_direction__02", "video_id": "uvo__4VZUCNBf1P4", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in a beige shirt and white cowboy hat riding a dark horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "right", "backward", "up"], "frame_paths": ["media/5aa7f5c08b926362a395/F0.jpg", "media/5aa7f5c08b926362a395/F1.jpg", "media/5aa7f5c08b926362a395/F2.jpg", "media/5aa7f5c08b926362a395/F3.jpg", "media/5aa7f5c08b926362a395/F4.jpg", "media/5aa7f5c08b926362a395/F5.jpg", "media/5aa7f5c08b926362a395/F6.jpg", "media/5aa7f5c08b926362a395/F7.jpg", "media/5aa7f5c08b926362a395/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in a beige shirt and white cowboy hat riding a dark horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. right\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "66fb8d2a7488456e58bc661d4b5ed2fe9ed961aeaf380997b7da9795046d2027"}
{"qa_id": "uvo__4VZUCNBf1P4__object_pair_distance_change__03", "video_id": "uvo__4VZUCNBf1P4", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, did the 3D distance between object A (person in a beige shirt and white cowboy hat riding a dark horse) and object B (person in a grey t-shirt riding a brown horse) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d7e4e49a271a91794b35/F0.jpg", "media/d7e4e49a271a91794b35/F1.jpg", "media/d7e4e49a271a91794b35/F2.jpg", "media/d7e4e49a271a91794b35/F3.jpg", "media/d7e4e49a271a91794b35/F4.jpg", "media/d7e4e49a271a91794b35/F5.jpg", "media/d7e4e49a271a91794b35/F6.jpg", "media/d7e4e49a271a91794b35/F7.jpg", "media/d7e4e49a271a91794b35/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did the 3D distance between object A (person in a beige shirt and white cowboy hat riding a dark horse) and object B (person in a grey t-shirt riding a brown horse) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "89cecad77853434c5fc3ccb545702776c7b362bd6b0fc9dc4afe27d403e15e7a"}
{"qa_id": "uvo__4Z6f8pFs-G4__object_world_motion_direction__00", "video_id": "uvo__4Z6f8pFs-G4", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (cross-country skier in orange bib) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "right", "forward", "left"], "frame_paths": ["media/1bca3a7663acb2181a3d/F0.jpg", "media/1bca3a7663acb2181a3d/F1.jpg", "media/1bca3a7663acb2181a3d/F2.jpg", "media/1bca3a7663acb2181a3d/F3.jpg", "media/1bca3a7663acb2181a3d/F4.jpg", "media/1bca3a7663acb2181a3d/F5.jpg", "media/1bca3a7663acb2181a3d/F6.jpg", "media/1bca3a7663acb2181a3d/F7.jpg", "media/1bca3a7663acb2181a3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (cross-country skier in orange bib) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e59a5c2df940e3c2d40e73f18ac169ec83e7051cddbdaf12e949140fc0a6f286"}
{"qa_id": "uvo__4Z6f8pFs-G4__camera_object_distance_change__01", "video_id": "uvo__4Z6f8pFs-G4", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F5, did object A (cross-country skier in orange bib) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/1bca3a7663acb2181a3d/F0.jpg", "media/1bca3a7663acb2181a3d/F1.jpg", "media/1bca3a7663acb2181a3d/F2.jpg", "media/1bca3a7663acb2181a3d/F3.jpg", "media/1bca3a7663acb2181a3d/F4.jpg", "media/1bca3a7663acb2181a3d/F5.jpg", "media/1bca3a7663acb2181a3d/F6.jpg", "media/1bca3a7663acb2181a3d/F7.jpg", "media/1bca3a7663acb2181a3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did object A (cross-country skier in orange bib) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "20a0a6f605b5d7428c296a41245101a273fd793dac408e6c6c5a7950a21cef77"}
{"qa_id": "uvo__4eRwcuIRiCg__object_world_motion_direction__00", "video_id": "uvo__4eRwcuIRiCg", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing ice skating) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "backward", "right", "down", "up"], "frame_paths": ["media/1c590b31e5d545cc42b8/F0.jpg", "media/1c590b31e5d545cc42b8/F1.jpg", "media/1c590b31e5d545cc42b8/F2.jpg", "media/1c590b31e5d545cc42b8/F3.jpg", "media/1c590b31e5d545cc42b8/F4.jpg", "media/1c590b31e5d545cc42b8/F5.jpg", "media/1c590b31e5d545cc42b8/F6.jpg", "media/1c590b31e5d545cc42b8/F7.jpg", "media/1c590b31e5d545cc42b8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in dark clothing ice skating) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7e7dbf28fddaa2becb09762434333b638b09e53cf0bccb0a052e296c01d32f8"}
{"qa_id": "uvo__4hwQbVmw4jo__object_world_motion_direction__00", "video_id": "uvo__4hwQbVmw4jo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (brown horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "backward", "right", "up"], "frame_paths": ["media/853a5a2b03361a74f30f/F0.jpg", "media/853a5a2b03361a74f30f/F1.jpg", "media/853a5a2b03361a74f30f/F2.jpg", "media/853a5a2b03361a74f30f/F3.jpg", "media/853a5a2b03361a74f30f/F4.jpg", "media/853a5a2b03361a74f30f/F5.jpg", "media/853a5a2b03361a74f30f/F6.jpg", "media/853a5a2b03361a74f30f/F7.jpg", "media/853a5a2b03361a74f30f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (brown horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. backward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87d151b6dd981eeae9118c494ab01c8f494febf7000578d2dc12ef1b79e1078a"}
{"qa_id": "uvo__4hwQbVmw4jo__object_world_motion_direction__01", "video_id": "uvo__4hwQbVmw4jo", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (green horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "right", "forward", "up", "down"], "frame_paths": ["media/f4a8e4b51485e577aa3d/F0.jpg", "media/f4a8e4b51485e577aa3d/F1.jpg", "media/f4a8e4b51485e577aa3d/F2.jpg", "media/f4a8e4b51485e577aa3d/F3.jpg", "media/f4a8e4b51485e577aa3d/F4.jpg", "media/f4a8e4b51485e577aa3d/F5.jpg", "media/f4a8e4b51485e577aa3d/F6.jpg", "media/f4a8e4b51485e577aa3d/F7.jpg", "media/f4a8e4b51485e577aa3d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (green horse) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f429112bf42c2466f534465f1cb46fb831371f29369d947b54ef30ef0611e1e0"}
{"qa_id": "uvo__4kjSczWgOS8__camera_object_distance_change__00", "video_id": "uvo__4kjSczWgOS8", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (brown horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/803de3ba251c6b14c4a2/F0.jpg", "media/803de3ba251c6b14c4a2/F1.jpg", "media/803de3ba251c6b14c4a2/F2.jpg", "media/803de3ba251c6b14c4a2/F3.jpg", "media/803de3ba251c6b14c4a2/F4.jpg", "media/803de3ba251c6b14c4a2/F5.jpg", "media/803de3ba251c6b14c4a2/F6.jpg", "media/803de3ba251c6b14c4a2/F7.jpg", "media/803de3ba251c6b14c4a2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (brown horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5856b9a41261e8e7d5920336f535dff1324141644c8e620b5b67c1c11254ecd"}
{"qa_id": "uvo__4kjSczWgOS8__camera_object_distance_change__01", "video_id": "uvo__4kjSczWgOS8", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (person wearing a white hat and blue shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/47f311181b2ba534ba37/F0.jpg", "media/47f311181b2ba534ba37/F1.jpg", "media/47f311181b2ba534ba37/F2.jpg", "media/47f311181b2ba534ba37/F3.jpg", "media/47f311181b2ba534ba37/F4.jpg", "media/47f311181b2ba534ba37/F5.jpg", "media/47f311181b2ba534ba37/F6.jpg", "media/47f311181b2ba534ba37/F7.jpg", "media/47f311181b2ba534ba37/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (person wearing a white hat and blue shirt riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "be0d24970c8dd192e3935e1f547bd743c9710231a2268e2157202337417009ae"}
{"qa_id": "uvo__4ul7FGez_V8__camera_translation_direction__00", "video_id": "uvo__4ul7FGez_V8", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "left", "right", "up", "backward"], "frame_paths": ["media/f948232a70569392035d/F0.jpg", "media/f948232a70569392035d/F1.jpg", "media/f948232a70569392035d/F2.jpg", "media/f948232a70569392035d/F3.jpg", "media/f948232a70569392035d/F4.jpg", "media/f948232a70569392035d/F5.jpg", "media/f948232a70569392035d/F6.jpg", "media/f948232a70569392035d/F7.jpg", "media/f948232a70569392035d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. left\nD. right\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "164615e5741e42a52cdb2bf895fcc42fbad6c2af7cdff9b90aaad1fb788ae925"}
{"qa_id": "uvo__52oamnB8PFc__camera_translation_direction__00", "video_id": "uvo__52oamnB8PFc", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "backward", "up", "forward"], "frame_paths": ["media/74b6651302760f7ceb4c/F0.jpg", "media/74b6651302760f7ceb4c/F1.jpg", "media/74b6651302760f7ceb4c/F2.jpg", "media/74b6651302760f7ceb4c/F3.jpg", "media/74b6651302760f7ceb4c/F4.jpg", "media/74b6651302760f7ceb4c/F5.jpg", "media/74b6651302760f7ceb4c/F6.jpg", "media/74b6651302760f7ceb4c/F7.jpg", "media/74b6651302760f7ceb4c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a909f21c8767038d61a5fcdab95fb3ac009c747adfef3266995b6230530a81f"}
{"qa_id": "uvo__54O-omRA7f8__camera_view_direction_change__00", "video_id": "uvo__54O-omRA7f8", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/a75e263504b992d76cda/F0.jpg", "media/a75e263504b992d76cda/F1.jpg", "media/a75e263504b992d76cda/F2.jpg", "media/a75e263504b992d76cda/F3.jpg", "media/a75e263504b992d76cda/F4.jpg", "media/a75e263504b992d76cda/F5.jpg", "media/a75e263504b992d76cda/F6.jpg", "media/a75e263504b992d76cda/F7.jpg", "media/a75e263504b992d76cda/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fac90f01284e0894afc97994f1a74f06180a57920337e6ab96aa8dfd96d817df"}
{"qa_id": "uvo__54O-omRA7f8__camera_object_distance_change__01", "video_id": "uvo__54O-omRA7f8", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (person in dark swimwear) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/e386f7b2a9d321ef432a/F0.jpg", "media/e386f7b2a9d321ef432a/F1.jpg", "media/e386f7b2a9d321ef432a/F2.jpg", "media/e386f7b2a9d321ef432a/F3.jpg", "media/e386f7b2a9d321ef432a/F4.jpg", "media/e386f7b2a9d321ef432a/F5.jpg", "media/e386f7b2a9d321ef432a/F6.jpg", "media/e386f7b2a9d321ef432a/F7.jpg", "media/e386f7b2a9d321ef432a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (person in dark swimwear) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f489eac295b95e7015e3fe6b00327ef18466a14b0c839edfdeaba0c191e6dea5"}
{"qa_id": "uvo__55rkeqdUVK4__camera_view_direction_change__00", "video_id": "uvo__55rkeqdUVK4", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "left", "right"], "frame_paths": ["media/1bc34b00d6abb07dd21d/F0.jpg", "media/1bc34b00d6abb07dd21d/F1.jpg", "media/1bc34b00d6abb07dd21d/F2.jpg", "media/1bc34b00d6abb07dd21d/F3.jpg", "media/1bc34b00d6abb07dd21d/F4.jpg", "media/1bc34b00d6abb07dd21d/F5.jpg", "media/1bc34b00d6abb07dd21d/F6.jpg", "media/1bc34b00d6abb07dd21d/F7.jpg", "media/1bc34b00d6abb07dd21d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c53e2b83a88fe2eec3cdec99b6f57c27ac0474645f750fa59a38cf69cf5470b3"}
{"qa_id": "uvo__5EZwqDdmBNY__camera_view_direction_change__00", "video_id": "uvo__5EZwqDdmBNY", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "up", "right", "left"], "frame_paths": ["media/95ca55f75b9290903ab1/F0.jpg", "media/95ca55f75b9290903ab1/F1.jpg", "media/95ca55f75b9290903ab1/F2.jpg", "media/95ca55f75b9290903ab1/F3.jpg", "media/95ca55f75b9290903ab1/F4.jpg", "media/95ca55f75b9290903ab1/F5.jpg", "media/95ca55f75b9290903ab1/F6.jpg", "media/95ca55f75b9290903ab1/F7.jpg", "media/95ca55f75b9290903ab1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. up\nC. right\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "139d78c609346a36b31a51baa6a9cfbf5b99d316434cbd3d4bd9ef34f83af385"}
{"qa_id": "uvo__5EZwqDdmBNY__object_world_motion_direction__01", "video_id": "uvo__5EZwqDdmBNY", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (person with long dark hair wearing a black coat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "forward", "up", "left"], "frame_paths": ["media/a92f4b6904fbf4e7c73e/F0.jpg", "media/a92f4b6904fbf4e7c73e/F1.jpg", "media/a92f4b6904fbf4e7c73e/F2.jpg", "media/a92f4b6904fbf4e7c73e/F3.jpg", "media/a92f4b6904fbf4e7c73e/F4.jpg", "media/a92f4b6904fbf4e7c73e/F5.jpg", "media/a92f4b6904fbf4e7c73e/F6.jpg", "media/a92f4b6904fbf4e7c73e/F7.jpg", "media/a92f4b6904fbf4e7c73e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (person with long dark hair wearing a black coat) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. forward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "45450527d36e270a612291ce186c7e1079ba743686fe2f7873bd70680a3e4012"}
{"qa_id": "uvo__5EZwqDdmBNY__object_travel_distance_comparison__02", "video_id": "uvo__5EZwqDdmBNY", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person with long dark hair wearing a black coat) or B (person with short grey hair wearing a dark blue jacket)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/8834b2f4209032c11f34/F0.jpg", "media/8834b2f4209032c11f34/F1.jpg", "media/8834b2f4209032c11f34/F2.jpg", "media/8834b2f4209032c11f34/F3.jpg", "media/8834b2f4209032c11f34/F4.jpg", "media/8834b2f4209032c11f34/F5.jpg", "media/8834b2f4209032c11f34/F6.jpg", "media/8834b2f4209032c11f34/F7.jpg", "media/8834b2f4209032c11f34/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g4"}, "requested_object_ids": ["g3", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person with long dark hair wearing a black coat) or B (person with short grey hair wearing a dark blue jacket)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eefc469ce400a69603fd54253d3c6748d66ac2b40a045e62b2cb444e15c6cb19"}
{"qa_id": "uvo__5IRgIqwt5MU__camera_view_direction_change__00", "video_id": "uvo__5IRgIqwt5MU", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/a8a08ce166f0b91ea04e/F0.jpg", "media/a8a08ce166f0b91ea04e/F1.jpg", "media/a8a08ce166f0b91ea04e/F2.jpg", "media/a8a08ce166f0b91ea04e/F3.jpg", "media/a8a08ce166f0b91ea04e/F4.jpg", "media/a8a08ce166f0b91ea04e/F5.jpg", "media/a8a08ce166f0b91ea04e/F6.jpg", "media/a8a08ce166f0b91ea04e/F7.jpg", "media/a8a08ce166f0b91ea04e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4be0070f398ed05c208baca3c0bbabd077ee5b3b77ffd593732c4fb964889062"}
{"qa_id": "uvo__5KF8Iz58290__object_world_motion_direction__00", "video_id": "uvo__5KF8Iz58290", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F7, in which dominant direction did object A (person in black t-shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "left", "backward", "forward"], "frame_paths": ["media/3c777d97beb11379ffab/F0.jpg", "media/3c777d97beb11379ffab/F1.jpg", "media/3c777d97beb11379ffab/F2.jpg", "media/3c777d97beb11379ffab/F3.jpg", "media/3c777d97beb11379ffab/F4.jpg", "media/3c777d97beb11379ffab/F5.jpg", "media/3c777d97beb11379ffab/F6.jpg", "media/3c777d97beb11379ffab/F7.jpg", "media/3c777d97beb11379ffab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F7, in which dominant direction did object A (person in black t-shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b2ff973428f6630a40f970d9847ae356598fc9b1681a523e96ba5007451e7d9"}
{"qa_id": "uvo__5PjPLbFbH_s__camera_translation_direction__00", "video_id": "uvo__5PjPLbFbH_s", "task": "camera_translation_direction", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "left", "right", "down"], "frame_paths": ["media/b82a18aa56e674a76d70/F0.jpg", "media/b82a18aa56e674a76d70/F1.jpg", "media/b82a18aa56e674a76d70/F2.jpg", "media/b82a18aa56e674a76d70/F3.jpg", "media/b82a18aa56e674a76d70/F4.jpg", "media/b82a18aa56e674a76d70/F5.jpg", "media/b82a18aa56e674a76d70/F6.jpg", "media/b82a18aa56e674a76d70/F7.jpg", "media/b82a18aa56e674a76d70/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. left\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a1da640da382579fa0f98b7598324dd5f4b1bf6bd1a420cddcca48b7a5e5fe2"}
{"qa_id": "uvo__5ZDcQMSeOUo__camera_view_direction_change__00", "video_id": "uvo__5ZDcQMSeOUo", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/a50b520cc13f33d3aa4f/F0.jpg", "media/a50b520cc13f33d3aa4f/F1.jpg", "media/a50b520cc13f33d3aa4f/F2.jpg", "media/a50b520cc13f33d3aa4f/F3.jpg", "media/a50b520cc13f33d3aa4f/F4.jpg", "media/a50b520cc13f33d3aa4f/F5.jpg", "media/a50b520cc13f33d3aa4f/F6.jpg", "media/a50b520cc13f33d3aa4f/F7.jpg", "media/a50b520cc13f33d3aa4f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2caa625aa254fd3abebce32193d4f99365c506ed6ff0451a16628a0cfc0281cc"}
{"qa_id": "uvo__5buTKoaekyI__object_travel_distance_comparison__00", "video_id": "uvo__5buTKoaekyI", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in light blue shirt) or B (woman in patterned blouse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/d15bcb90a1419d398085/F0.jpg", "media/d15bcb90a1419d398085/F1.jpg", "media/d15bcb90a1419d398085/F2.jpg", "media/d15bcb90a1419d398085/F3.jpg", "media/d15bcb90a1419d398085/F4.jpg", "media/d15bcb90a1419d398085/F5.jpg", "media/d15bcb90a1419d398085/F6.jpg", "media/d15bcb90a1419d398085/F7.jpg", "media/d15bcb90a1419d398085/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (man in light blue shirt) or B (woman in patterned blouse)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2b550affb15a31a9971b1a8b3cbc9522a5e1bb4b643836e7e8593948b8562a6c"}
{"qa_id": "uvo__5hCBTx6i9Rk__object_pair_distance_change__00", "video_id": "uvo__5hCBTx6i9Rk", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F5, did the 3D distance between object A (hockey player in red jersey) and object B (hockey player in white jersey) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/c1c7577415806be2f88d/F0.jpg", "media/c1c7577415806be2f88d/F1.jpg", "media/c1c7577415806be2f88d/F2.jpg", "media/c1c7577415806be2f88d/F3.jpg", "media/c1c7577415806be2f88d/F4.jpg", "media/c1c7577415806be2f88d/F5.jpg", "media/c1c7577415806be2f88d/F6.jpg", "media/c1c7577415806be2f88d/F7.jpg", "media/c1c7577415806be2f88d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, did the 3D distance between object A (hockey player in red jersey) and object B (hockey player in white jersey) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "257f27adb97d0b4c938cc9f92d1bee705efea56230fcf1dd508b9348d6d0a231"}
{"qa_id": "uvo__5hHJPNxhPAo__object_world_motion_direction__00", "video_id": "uvo__5hHJPNxhPAo", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (green tractor with a person in a yellow shirt driving) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "left", "down", "forward", "right"], "frame_paths": ["media/10663683338e8de3dd5a/F0.jpg", "media/10663683338e8de3dd5a/F1.jpg", "media/10663683338e8de3dd5a/F2.jpg", "media/10663683338e8de3dd5a/F3.jpg", "media/10663683338e8de3dd5a/F4.jpg", "media/10663683338e8de3dd5a/F5.jpg", "media/10663683338e8de3dd5a/F6.jpg", "media/10663683338e8de3dd5a/F7.jpg", "media/10663683338e8de3dd5a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (green tractor with a person in a yellow shirt driving) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. left\nD. down\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa7b4db0b12d73dc39ac5ece317d9cd9daa9d8c6cd048b5aeb47c45ebc8b4690"}
{"qa_id": "uvo__5unu_IlrT7E__camera_view_direction_change__00", "video_id": "uvo__5unu_IlrT7E", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/c223c7ea19653ceaa3d4/F0.jpg", "media/c223c7ea19653ceaa3d4/F1.jpg", "media/c223c7ea19653ceaa3d4/F2.jpg", "media/c223c7ea19653ceaa3d4/F3.jpg", "media/c223c7ea19653ceaa3d4/F4.jpg", "media/c223c7ea19653ceaa3d4/F5.jpg", "media/c223c7ea19653ceaa3d4/F6.jpg", "media/c223c7ea19653ceaa3d4/F7.jpg", "media/c223c7ea19653ceaa3d4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9e9b76767acadd54a6de2adfd760c5c5c5a40f6713b7b4e384d9c5a0b3abaa4"}
{"qa_id": "uvo__Dm718Ltyt2Q__object_pair_distance_change__00", "video_id": "uvo__Dm718Ltyt2Q", "task": "object_pair_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F6, did the 3D distance between object A (person in orange shirt) and object B (person in green shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/1a5743c3b2fa921c84b6/F0.jpg", "media/1a5743c3b2fa921c84b6/F1.jpg", "media/1a5743c3b2fa921c84b6/F2.jpg", "media/1a5743c3b2fa921c84b6/F3.jpg", "media/1a5743c3b2fa921c84b6/F4.jpg", "media/1a5743c3b2fa921c84b6/F5.jpg", "media/1a5743c3b2fa921c84b6/F6.jpg", "media/1a5743c3b2fa921c84b6/F7.jpg", "media/1a5743c3b2fa921c84b6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did the 3D distance between object A (person in orange shirt) and object B (person in green shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c0138d77f0f24dd39f4b3c098637ae799d55fc084854ed4ebb83b2586fee1945"}
{"qa_id": "uvo__Dm718Ltyt2Q__object_travel_distance_comparison__01", "video_id": "uvo__Dm718Ltyt2Q", "task": "object_travel_distance_comparison", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt) or B (person in green shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/1a5743c3b2fa921c84b6/F0.jpg", "media/1a5743c3b2fa921c84b6/F1.jpg", "media/1a5743c3b2fa921c84b6/F2.jpg", "media/1a5743c3b2fa921c84b6/F3.jpg", "media/1a5743c3b2fa921c84b6/F4.jpg", "media/1a5743c3b2fa921c84b6/F5.jpg", "media/1a5743c3b2fa921c84b6/F6.jpg", "media/1a5743c3b2fa921c84b6/F7.jpg", "media/1a5743c3b2fa921c84b6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F1 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt) or B (person in green shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5c1d3c399228c4f852a640b012b9ead2c5dd1b371902e8254bbc93470084adbc"}
{"qa_id": "uvo__GRDtDxRR1uE__camera_translation_direction__00", "video_id": "uvo__GRDtDxRR1uE", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "up", "down", "right", "forward"], "frame_paths": ["media/72303b3ca0f23c338ea5/F0.jpg", "media/72303b3ca0f23c338ea5/F1.jpg", "media/72303b3ca0f23c338ea5/F2.jpg", "media/72303b3ca0f23c338ea5/F3.jpg", "media/72303b3ca0f23c338ea5/F4.jpg", "media/72303b3ca0f23c338ea5/F5.jpg", "media/72303b3ca0f23c338ea5/F6.jpg", "media/72303b3ca0f23c338ea5/F7.jpg", "media/72303b3ca0f23c338ea5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. up\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "08fc2a3ed3de539937c692a935045fb1827a9211024eb3a2f8710e99c8dc4b09"}
{"qa_id": "uvo__GRDtDxRR1uE__camera_object_distance_change__01", "video_id": "uvo__GRDtDxRR1uE", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object A (toddler in a light purple sleeveless top and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/998a2527b677a9159aa4/F0.jpg", "media/998a2527b677a9159aa4/F1.jpg", "media/998a2527b677a9159aa4/F2.jpg", "media/998a2527b677a9159aa4/F3.jpg", "media/998a2527b677a9159aa4/F4.jpg", "media/998a2527b677a9159aa4/F5.jpg", "media/998a2527b677a9159aa4/F6.jpg", "media/998a2527b677a9159aa4/F7.jpg", "media/998a2527b677a9159aa4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (toddler in a light purple sleeveless top and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a75c89f0154e2465c1ba40751ef098c88d5b3e0c8baf8b7b40e8d5f5d6e2cfc2"}
{"qa_id": "uvo__GRDtDxRR1uE__camera_object_distance_change__02", "video_id": "uvo__GRDtDxRR1uE", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did object B (light purple ride-on toy with a handle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/db0ccbdca3084dae0615/F0.jpg", "media/db0ccbdca3084dae0615/F1.jpg", "media/db0ccbdca3084dae0615/F2.jpg", "media/db0ccbdca3084dae0615/F3.jpg", "media/db0ccbdca3084dae0615/F4.jpg", "media/db0ccbdca3084dae0615/F5.jpg", "media/db0ccbdca3084dae0615/F6.jpg", "media/db0ccbdca3084dae0615/F7.jpg", "media/db0ccbdca3084dae0615/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (light purple ride-on toy with a handle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "19a2e8e3c0dc6f8c7d90fbd0d2e37a9f851e808e16e269e7598f23449330d839"}
{"qa_id": "uvo__GXbWyROGNa8__camera_view_direction_change__00", "video_id": "uvo__GXbWyROGNa8", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/e3c224fce4f69c3fe368/F0.jpg", "media/e3c224fce4f69c3fe368/F1.jpg", "media/e3c224fce4f69c3fe368/F2.jpg", "media/e3c224fce4f69c3fe368/F3.jpg", "media/e3c224fce4f69c3fe368/F4.jpg", "media/e3c224fce4f69c3fe368/F5.jpg", "media/e3c224fce4f69c3fe368/F6.jpg", "media/e3c224fce4f69c3fe368/F7.jpg", "media/e3c224fce4f69c3fe368/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bef315086a2e22e0c62fbd5310f222503d99f8f4104d43b230d064bf174c4882"}
{"qa_id": "uvo__GbYJPAXci50__object_world_motion_direction__00", "video_id": "uvo__GbYJPAXci50", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "backward", "up", "forward"], "frame_paths": ["media/2dfea14cba964089bb89/F0.jpg", "media/2dfea14cba964089bb89/F1.jpg", "media/2dfea14cba964089bb89/F2.jpg", "media/2dfea14cba964089bb89/F3.jpg", "media/2dfea14cba964089bb89/F4.jpg", "media/2dfea14cba964089bb89/F5.jpg", "media/2dfea14cba964089bb89/F6.jpg", "media/2dfea14cba964089bb89/F7.jpg", "media/2dfea14cba964089bb89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3caad2b25e9fdf1ec7759f7d8aca588a84609801a7362bed534850dfaee1ee5c"}
{"qa_id": "uvo__GbYJPAXci50__object_world_motion_direction__01", "video_id": "uvo__GbYJPAXci50", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object B (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "right", "forward", "backward", "down", "up"], "frame_paths": ["media/fda636a5536364371807/F0.jpg", "media/fda636a5536364371807/F1.jpg", "media/fda636a5536364371807/F2.jpg", "media/fda636a5536364371807/F3.jpg", "media/fda636a5536364371807/F4.jpg", "media/fda636a5536364371807/F5.jpg", "media/fda636a5536364371807/F6.jpg", "media/fda636a5536364371807/F7.jpg", "media/fda636a5536364371807/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object B (white sedan) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. forward\nD. backward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ea12e1aec1c1e3e7e2a2c94b573c6872ffc770a47a288597b7daef7f62c5489b"}
{"qa_id": "uvo__GbYJPAXci50__object_pair_distance_change__02", "video_id": "uvo__GbYJPAXci50", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (white sedan) and object B (white sedan) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/29102d5e9678fbe2d51e/F0.jpg", "media/29102d5e9678fbe2d51e/F1.jpg", "media/29102d5e9678fbe2d51e/F2.jpg", "media/29102d5e9678fbe2d51e/F3.jpg", "media/29102d5e9678fbe2d51e/F4.jpg", "media/29102d5e9678fbe2d51e/F5.jpg", "media/29102d5e9678fbe2d51e/F6.jpg", "media/29102d5e9678fbe2d51e/F7.jpg", "media/29102d5e9678fbe2d51e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (white sedan) and object B (white sedan) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "298726eb4433c96e2bd196c9d55b91efa8f5debfbcc3f6498442bb1befe57b92"}
{"qa_id": "uvo__GbYJPAXci50__object_travel_distance_comparison__03", "video_id": "uvo__GbYJPAXci50", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white sedan) or B (white sedan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/29102d5e9678fbe2d51e/F0.jpg", "media/29102d5e9678fbe2d51e/F1.jpg", "media/29102d5e9678fbe2d51e/F2.jpg", "media/29102d5e9678fbe2d51e/F3.jpg", "media/29102d5e9678fbe2d51e/F4.jpg", "media/29102d5e9678fbe2d51e/F5.jpg", "media/29102d5e9678fbe2d51e/F6.jpg", "media/29102d5e9678fbe2d51e/F7.jpg", "media/29102d5e9678fbe2d51e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white sedan) or B (white sedan)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "039c09943dbe46d27c11388db9772ef34d2a24d9ca1bc3d2aa6533351c82fd6a"}
{"qa_id": "uvo__GydUCcBmL_g__camera_object_distance_change__01", "video_id": "uvo__GydUCcBmL_g", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object A (black car partially visible on the left edge) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d45b742c387bc3ab3630/F0.jpg", "media/d45b742c387bc3ab3630/F1.jpg", "media/d45b742c387bc3ab3630/F2.jpg", "media/d45b742c387bc3ab3630/F3.jpg", "media/d45b742c387bc3ab3630/F4.jpg", "media/d45b742c387bc3ab3630/F5.jpg", "media/d45b742c387bc3ab3630/F6.jpg", "media/d45b742c387bc3ab3630/F7.jpg", "media/d45b742c387bc3ab3630/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (black car partially visible on the left edge) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b9fc4c03051ee5f790a82b36ab9c38307d2cb646a339c3d42c64f9af906d1ec9"}
{"qa_id": "uvo__GydUCcBmL_g__object_world_motion_direction__02", "video_id": "uvo__GydUCcBmL_g", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object B (white car parked on the side of the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "left", "backward", "up", "down"], "frame_paths": ["media/8748867e2200ee03743a/F0.jpg", "media/8748867e2200ee03743a/F1.jpg", "media/8748867e2200ee03743a/F2.jpg", "media/8748867e2200ee03743a/F3.jpg", "media/8748867e2200ee03743a/F4.jpg", "media/8748867e2200ee03743a/F5.jpg", "media/8748867e2200ee03743a/F6.jpg", "media/8748867e2200ee03743a/F7.jpg", "media/8748867e2200ee03743a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object B (white car parked on the side of the road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. left\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0e72cc9b087e9d03b18ee507f0ebdc862011e1020ac5d497511eafa4beead471"}
{"qa_id": "uvo__GydUCcBmL_g__object_pair_distance_change__03", "video_id": "uvo__GydUCcBmL_g", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did the 3D distance between object A (black car partially visible on the left edge) and object B (white car parked on the side of the road) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/7795f80241b332e56598/F0.jpg", "media/7795f80241b332e56598/F1.jpg", "media/7795f80241b332e56598/F2.jpg", "media/7795f80241b332e56598/F3.jpg", "media/7795f80241b332e56598/F4.jpg", "media/7795f80241b332e56598/F5.jpg", "media/7795f80241b332e56598/F6.jpg", "media/7795f80241b332e56598/F7.jpg", "media/7795f80241b332e56598/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did the 3D distance between object A (black car partially visible on the left edge) and object B (white car parked on the side of the road) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0af8de849dc9d91a540c596d3192ed0491c6f1e92c4084bd937a6cb87dfc5464"}
{"qa_id": "uvo__NMKc_U02sxY__camera_translation_direction__00", "video_id": "uvo__NMKc_U02sxY", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "right", "up", "forward"], "frame_paths": ["media/8b8cf7ec2db37013e6d2/F0.jpg", "media/8b8cf7ec2db37013e6d2/F1.jpg", "media/8b8cf7ec2db37013e6d2/F2.jpg", "media/8b8cf7ec2db37013e6d2/F3.jpg", "media/8b8cf7ec2db37013e6d2/F4.jpg", "media/8b8cf7ec2db37013e6d2/F5.jpg", "media/8b8cf7ec2db37013e6d2/F6.jpg", "media/8b8cf7ec2db37013e6d2/F7.jpg", "media/8b8cf7ec2db37013e6d2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "17bf48e54965dca3941f0c193ee140fd0c381a9c643a0d9c75a55c140ace33e7"}
{"qa_id": "uvo__NsBvMqIDGZs__camera_view_direction_change__00", "video_id": "uvo__NsBvMqIDGZs", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/b34a41376bf553e55a74/F0.jpg", "media/b34a41376bf553e55a74/F1.jpg", "media/b34a41376bf553e55a74/F2.jpg", "media/b34a41376bf553e55a74/F3.jpg", "media/b34a41376bf553e55a74/F4.jpg", "media/b34a41376bf553e55a74/F5.jpg", "media/b34a41376bf553e55a74/F6.jpg", "media/b34a41376bf553e55a74/F7.jpg", "media/b34a41376bf553e55a74/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d74fca09116d39d356a6ef2fd8b0c080f84910eca3a92d7cbacd4bbec359d61c"}
{"qa_id": "uvo__ehKx69TyL5Q__camera_view_direction_change__00", "video_id": "uvo__ehKx69TyL5Q", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/1d9402684d6c1397fbe1/F0.jpg", "media/1d9402684d6c1397fbe1/F1.jpg", "media/1d9402684d6c1397fbe1/F2.jpg", "media/1d9402684d6c1397fbe1/F3.jpg", "media/1d9402684d6c1397fbe1/F4.jpg", "media/1d9402684d6c1397fbe1/F5.jpg", "media/1d9402684d6c1397fbe1/F6.jpg", "media/1d9402684d6c1397fbe1/F7.jpg", "media/1d9402684d6c1397fbe1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8f3290b3246a69f21031c92bcd8f3354b9d9ed5d12e9057d6a997ac7f477ae73"}
{"qa_id": "uvo__ehKx69TyL5Q__object_world_motion_direction__01", "video_id": "uvo__ehKx69TyL5Q", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object B (person hanging from a bungee cord) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "backward", "left", "up"], "frame_paths": ["media/38a4cc355e16a72f726f/F0.jpg", "media/38a4cc355e16a72f726f/F1.jpg", "media/38a4cc355e16a72f726f/F2.jpg", "media/38a4cc355e16a72f726f/F3.jpg", "media/38a4cc355e16a72f726f/F4.jpg", "media/38a4cc355e16a72f726f/F5.jpg", "media/38a4cc355e16a72f726f/F6.jpg", "media/38a4cc355e16a72f726f/F7.jpg", "media/38a4cc355e16a72f726f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object B (person hanging from a bungee cord) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. backward\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84d3b9659a9ff1825fa8ae516977b198e7973874ee8d6a0e2b70046c20aa9e77"}
{"qa_id": "uvo__ehKx69TyL5Q__object_pair_distance_change__02", "video_id": "uvo__ehKx69TyL5Q", "task": "object_pair_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, did the 3D distance between object A (person in orange shirt on the platform) and object B (person hanging from a bungee cord) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/15af0590d3bddde60fb3/F0.jpg", "media/15af0590d3bddde60fb3/F1.jpg", "media/15af0590d3bddde60fb3/F2.jpg", "media/15af0590d3bddde60fb3/F3.jpg", "media/15af0590d3bddde60fb3/F4.jpg", "media/15af0590d3bddde60fb3/F5.jpg", "media/15af0590d3bddde60fb3/F6.jpg", "media/15af0590d3bddde60fb3/F7.jpg", "media/15af0590d3bddde60fb3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did the 3D distance between object A (person in orange shirt on the platform) and object B (person hanging from a bungee cord) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "52cb7cca0b6e73f29a1ec43eb24dd96179058645d9799c41ccfb858be5a8b854"}
{"qa_id": "uvo__ehKx69TyL5Q__object_travel_distance_comparison__03", "video_id": "uvo__ehKx69TyL5Q", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt on the platform) or B (person hanging from a bungee cord)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/15af0590d3bddde60fb3/F0.jpg", "media/15af0590d3bddde60fb3/F1.jpg", "media/15af0590d3bddde60fb3/F2.jpg", "media/15af0590d3bddde60fb3/F3.jpg", "media/15af0590d3bddde60fb3/F4.jpg", "media/15af0590d3bddde60fb3/F5.jpg", "media/15af0590d3bddde60fb3/F6.jpg", "media/15af0590d3bddde60fb3/F7.jpg", "media/15af0590d3bddde60fb3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F6 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in orange shirt on the platform) or B (person hanging from a bungee cord)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "afea21b6238b1e79ad14c57b81dd21a7efde386ecbf08b0c5e28c82d8e1501e1"}
{"qa_id": "uvo__gB3ZQCg9XtM__camera_translation_direction__00", "video_id": "uvo__gB3ZQCg9XtM", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "forward", "up", "down"], "frame_paths": ["media/a51a6e5e3a1fdcf6c70f/F0.jpg", "media/a51a6e5e3a1fdcf6c70f/F1.jpg", "media/a51a6e5e3a1fdcf6c70f/F2.jpg", "media/a51a6e5e3a1fdcf6c70f/F3.jpg", "media/a51a6e5e3a1fdcf6c70f/F4.jpg", "media/a51a6e5e3a1fdcf6c70f/F5.jpg", "media/a51a6e5e3a1fdcf6c70f/F6.jpg", "media/a51a6e5e3a1fdcf6c70f/F7.jpg", "media/a51a6e5e3a1fdcf6c70f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "328fe67b612bace3d9013d617c9b18cc1d290e1a2b790a30c6d46264d6c5fed6"}
{"qa_id": "uvo__nUEZNKi_98Y__camera_translation_direction__00", "video_id": "uvo__nUEZNKi_98Y", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "left", "forward", "backward"], "frame_paths": ["media/240abb8a2f6449c4aa7d/F0.jpg", "media/240abb8a2f6449c4aa7d/F1.jpg", "media/240abb8a2f6449c4aa7d/F2.jpg", "media/240abb8a2f6449c4aa7d/F3.jpg", "media/240abb8a2f6449c4aa7d/F4.jpg", "media/240abb8a2f6449c4aa7d/F5.jpg", "media/240abb8a2f6449c4aa7d/F6.jpg", "media/240abb8a2f6449c4aa7d/F7.jpg", "media/240abb8a2f6449c4aa7d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. left\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ddd4e0487433bdc88d7b27723839363d50a8f3d77ba16b9fc320e468b2641a7f"}
{"qa_id": "uvo__nUEZNKi_98Y__object_world_motion_direction__01", "video_id": "uvo__nUEZNKi_98Y", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person water skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/4248e7240345d2dcbf40/F0.jpg", "media/4248e7240345d2dcbf40/F1.jpg", "media/4248e7240345d2dcbf40/F2.jpg", "media/4248e7240345d2dcbf40/F3.jpg", "media/4248e7240345d2dcbf40/F4.jpg", "media/4248e7240345d2dcbf40/F5.jpg", "media/4248e7240345d2dcbf40/F6.jpg", "media/4248e7240345d2dcbf40/F7.jpg", "media/4248e7240345d2dcbf40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person water skiing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d90ffd71213403d7112e0379bc701a5cb8a48eb650b885f7507b74c02f464717"}
{"qa_id": "uvo__nUEZNKi_98Y__camera_object_distance_change__02", "video_id": "uvo__nUEZNKi_98Y", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (person water skiing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/4248e7240345d2dcbf40/F0.jpg", "media/4248e7240345d2dcbf40/F1.jpg", "media/4248e7240345d2dcbf40/F2.jpg", "media/4248e7240345d2dcbf40/F3.jpg", "media/4248e7240345d2dcbf40/F4.jpg", "media/4248e7240345d2dcbf40/F5.jpg", "media/4248e7240345d2dcbf40/F6.jpg", "media/4248e7240345d2dcbf40/F7.jpg", "media/4248e7240345d2dcbf40/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person water skiing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "86b21e5253606731542c9f1ca5e32125aef4bef16330344fbd1385bd98d1bfe6"}
{"qa_id": "uvo__nUEZNKi_98Y__object_world_motion_direction__03", "video_id": "uvo__nUEZNKi_98Y", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (white water ski) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "left", "up", "backward"], "frame_paths": ["media/7ca641a3f8899e2f1795/F0.jpg", "media/7ca641a3f8899e2f1795/F1.jpg", "media/7ca641a3f8899e2f1795/F2.jpg", "media/7ca641a3f8899e2f1795/F3.jpg", "media/7ca641a3f8899e2f1795/F4.jpg", "media/7ca641a3f8899e2f1795/F5.jpg", "media/7ca641a3f8899e2f1795/F6.jpg", "media/7ca641a3f8899e2f1795/F7.jpg", "media/7ca641a3f8899e2f1795/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (white water ski) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f06034090f2ed515317ca24ae955ae1fd459985190bbb4d2224d3bc88f04cc41"}
{"qa_id": "uvo__nUEZNKi_98Y__camera_object_distance_change__04", "video_id": "uvo__nUEZNKi_98Y", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (white water ski) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7ca641a3f8899e2f1795/F0.jpg", "media/7ca641a3f8899e2f1795/F1.jpg", "media/7ca641a3f8899e2f1795/F2.jpg", "media/7ca641a3f8899e2f1795/F3.jpg", "media/7ca641a3f8899e2f1795/F4.jpg", "media/7ca641a3f8899e2f1795/F5.jpg", "media/7ca641a3f8899e2f1795/F6.jpg", "media/7ca641a3f8899e2f1795/F7.jpg", "media/7ca641a3f8899e2f1795/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (white water ski) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9a1f48559f078095024d87bf10eee981a4c3d68b9b39f6f56b50bd751312726d"}
{"qa_id": "uvo__p7N2efIPu10__camera_translation_direction__00", "video_id": "uvo__p7N2efIPu10", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "up", "down"], "frame_paths": ["media/bda21ec09cc1869b41f3/F0.jpg", "media/bda21ec09cc1869b41f3/F1.jpg", "media/bda21ec09cc1869b41f3/F2.jpg", "media/bda21ec09cc1869b41f3/F3.jpg", "media/bda21ec09cc1869b41f3/F4.jpg", "media/bda21ec09cc1869b41f3/F5.jpg", "media/bda21ec09cc1869b41f3/F6.jpg", "media/bda21ec09cc1869b41f3/F7.jpg", "media/bda21ec09cc1869b41f3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d416d1c7501c6012aa162c3db417fc26f51448ee531be03644fab0c8a9f69cda"}
{"qa_id": "uvo__xrygOElnq5I__camera_view_direction_change__00", "video_id": "uvo__xrygOElnq5I", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/c763771a2c4606766ad9/F0.jpg", "media/c763771a2c4606766ad9/F1.jpg", "media/c763771a2c4606766ad9/F2.jpg", "media/c763771a2c4606766ad9/F3.jpg", "media/c763771a2c4606766ad9/F4.jpg", "media/c763771a2c4606766ad9/F5.jpg", "media/c763771a2c4606766ad9/F6.jpg", "media/c763771a2c4606766ad9/F7.jpg", "media/c763771a2c4606766ad9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "uvo__xrygOElnq5I__camera_object_distance_change__01", "video_id": "uvo__xrygOElnq5I", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, did object A (person in white shirt and orange cap) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/5f860bc274cfde134436/F0.jpg", "media/5f860bc274cfde134436/F1.jpg", "media/5f860bc274cfde134436/F2.jpg", "media/5f860bc274cfde134436/F3.jpg", "media/5f860bc274cfde134436/F4.jpg", "media/5f860bc274cfde134436/F5.jpg", "media/5f860bc274cfde134436/F6.jpg", "media/5f860bc274cfde134436/F7.jpg", "media/5f860bc274cfde134436/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (person in white shirt and orange cap) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "95bd3330687747359f658ba9a534b4b1a5040082b24dc4b054dbd978a1cd5b47"}
{"qa_id": "youtube_vis__019cf5e661__object_world_motion_direction__00", "video_id": "youtube_vis__019cf5e661", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (large adult polar bear walking on the rock ledge) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "right", "backward", "forward", "down", "up"], "frame_paths": ["media/20becd97db47040f153c/F0.jpg", "media/20becd97db47040f153c/F1.jpg", "media/20becd97db47040f153c/F2.jpg", "media/20becd97db47040f153c/F3.jpg", "media/20becd97db47040f153c/F4.jpg", "media/20becd97db47040f153c/F5.jpg", "media/20becd97db47040f153c/F6.jpg", "media/20becd97db47040f153c/F7.jpg", "media/20becd97db47040f153c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (large adult polar bear walking on the rock ledge) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. backward\nD. forward\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f7fabce83f9f359bc39f6d0534df279a14aae22bcf735eceaab8c7e9d302b98d"}
{"qa_id": "youtube_vis__03fe889119__camera_translation_direction__00", "video_id": "youtube_vis__03fe889119", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "left", "up", "backward"], "frame_paths": ["media/109ab35210a988b6096e/F0.jpg", "media/109ab35210a988b6096e/F1.jpg", "media/109ab35210a988b6096e/F2.jpg", "media/109ab35210a988b6096e/F3.jpg", "media/109ab35210a988b6096e/F4.jpg", "media/109ab35210a988b6096e/F5.jpg", "media/109ab35210a988b6096e/F6.jpg", "media/109ab35210a988b6096e/F7.jpg", "media/109ab35210a988b6096e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5943a514ec79590b682661a109ecce8606cd932ea0138de00a85a9914543161"}
{"qa_id": "youtube_vis__03fe889119__object_world_motion_direction__01", "video_id": "youtube_vis__03fe889119", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (red sedan) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "right", "forward", "up", "down"], "frame_paths": ["media/f99f6ff537c9a3078215/F0.jpg", "media/f99f6ff537c9a3078215/F1.jpg", "media/f99f6ff537c9a3078215/F2.jpg", "media/f99f6ff537c9a3078215/F3.jpg", "media/f99f6ff537c9a3078215/F4.jpg", "media/f99f6ff537c9a3078215/F5.jpg", "media/f99f6ff537c9a3078215/F6.jpg", "media/f99f6ff537c9a3078215/F7.jpg", "media/f99f6ff537c9a3078215/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (red sedan) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. right\nD. forward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05b9f94db40f8681ba9830d018a3b84eab610520770d8f05bf7f9d3e730e232b"}
{"qa_id": "youtube_vis__04d62d9d98__object_world_motion_direction__00", "video_id": "youtube_vis__04d62d9d98", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in white t-shirt and black pants skateboarding) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "down", "left", "forward"], "frame_paths": ["media/f7fb9333fdfb747f031b/F0.jpg", "media/f7fb9333fdfb747f031b/F1.jpg", "media/f7fb9333fdfb747f031b/F2.jpg", "media/f7fb9333fdfb747f031b/F3.jpg", "media/f7fb9333fdfb747f031b/F4.jpg", "media/f7fb9333fdfb747f031b/F5.jpg", "media/f7fb9333fdfb747f031b/F6.jpg", "media/f7fb9333fdfb747f031b/F7.jpg", "media/f7fb9333fdfb747f031b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in white t-shirt and black pants skateboarding) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6a26f93b96f3c5f218472669e63d65b2ace23a6a68de3f231680a22469d412b6"}
{"qa_id": "youtube_vis__05a0a513df__camera_object_distance_change__00", "video_id": "youtube_vis__05a0a513df", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, did object A (person in a dark jacket) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/f3a27323925080baf857/F0.jpg", "media/f3a27323925080baf857/F1.jpg", "media/f3a27323925080baf857/F2.jpg", "media/f3a27323925080baf857/F3.jpg", "media/f3a27323925080baf857/F4.jpg", "media/f3a27323925080baf857/F5.jpg", "media/f3a27323925080baf857/F6.jpg", "media/f3a27323925080baf857/F7.jpg", "media/f3a27323925080baf857/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (person in a dark jacket) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "05cefaff52217853630448ea8bffd1be58ec9874299f084c0193fdd09f10c3c9"}
{"qa_id": "youtube_vis__0630391881__camera_object_distance_change__00", "video_id": "youtube_vis__0630391881", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F7, did object A (person in black shirt and red shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/38e274d48ca5c30ed1f3/F0.jpg", "media/38e274d48ca5c30ed1f3/F1.jpg", "media/38e274d48ca5c30ed1f3/F2.jpg", "media/38e274d48ca5c30ed1f3/F3.jpg", "media/38e274d48ca5c30ed1f3/F4.jpg", "media/38e274d48ca5c30ed1f3/F5.jpg", "media/38e274d48ca5c30ed1f3/F6.jpg", "media/38e274d48ca5c30ed1f3/F7.jpg", "media/38e274d48ca5c30ed1f3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person in black shirt and red shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f54dc351fba9c4cc542603510a6ca5a9e6a00cd9200239b9c2d3d511a0a0c440"}
{"qa_id": "youtube_vis__084c70ced0__camera_translation_direction__00", "video_id": "youtube_vis__084c70ced0", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "left", "down", "up", "right"], "frame_paths": ["media/564b05b285c7e668016b/F0.jpg", "media/564b05b285c7e668016b/F1.jpg", "media/564b05b285c7e668016b/F2.jpg", "media/564b05b285c7e668016b/F3.jpg", "media/564b05b285c7e668016b/F4.jpg", "media/564b05b285c7e668016b/F5.jpg", "media/564b05b285c7e668016b/F6.jpg", "media/564b05b285c7e668016b/F7.jpg", "media/564b05b285c7e668016b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. left\nD. down\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c03794453986f7d01b34bb1531b4c452509e9713eee8e33ab9c124ffa4de283c"}
{"qa_id": "youtube_vis__084c70ced0__camera_view_direction_change__01", "video_id": "youtube_vis__084c70ced0", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/564b05b285c7e668016b/F0.jpg", "media/564b05b285c7e668016b/F1.jpg", "media/564b05b285c7e668016b/F2.jpg", "media/564b05b285c7e668016b/F3.jpg", "media/564b05b285c7e668016b/F4.jpg", "media/564b05b285c7e668016b/F5.jpg", "media/564b05b285c7e668016b/F6.jpg", "media/564b05b285c7e668016b/F7.jpg", "media/564b05b285c7e668016b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a5c501d1ea2df63daa56f4e051eddbbe4a5e28a274c0a5194d49800aef410114"}
{"qa_id": "youtube_vis__084c70ced0__object_world_motion_direction__02", "video_id": "youtube_vis__084c70ced0", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (man in a white hat and green waders holding a fishing rod) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "forward", "left", "up", "right"], "frame_paths": ["media/d7c20e4d0f963e1862dd/F0.jpg", "media/d7c20e4d0f963e1862dd/F1.jpg", "media/d7c20e4d0f963e1862dd/F2.jpg", "media/d7c20e4d0f963e1862dd/F3.jpg", "media/d7c20e4d0f963e1862dd/F4.jpg", "media/d7c20e4d0f963e1862dd/F5.jpg", "media/d7c20e4d0f963e1862dd/F6.jpg", "media/d7c20e4d0f963e1862dd/F7.jpg", "media/d7c20e4d0f963e1862dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (man in a white hat and green waders holding a fishing rod) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. forward\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca510d08a6c6359a9cbc7e382e675e48dd87a0e6429b1075e2d0842b0432831a"}
{"qa_id": "youtube_vis__088c295569__camera_view_direction_change__00", "video_id": "youtube_vis__088c295569", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["right", "left", "down", "up"], "frame_paths": ["media/00ad5a37c6080152b0f0/F0.jpg", "media/00ad5a37c6080152b0f0/F1.jpg", "media/00ad5a37c6080152b0f0/F2.jpg", "media/00ad5a37c6080152b0f0/F3.jpg", "media/00ad5a37c6080152b0f0/F4.jpg", "media/00ad5a37c6080152b0f0/F5.jpg", "media/00ad5a37c6080152b0f0/F6.jpg", "media/00ad5a37c6080152b0f0/F7.jpg", "media/00ad5a37c6080152b0f0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. right\nB. left\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5683ac2cba2889761f2107cb0cd0e4049566bdfcb4d4d2e9e175afe1d3d5de8"}
{"qa_id": "youtube_vis__088c295569__object_world_motion_direction__01", "video_id": "youtube_vis__088c295569", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (white rally car with orange number 91) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "down", "up", "right", "left", "forward"], "frame_paths": ["media/b4635dee01d767cefaa7/F0.jpg", "media/b4635dee01d767cefaa7/F1.jpg", "media/b4635dee01d767cefaa7/F2.jpg", "media/b4635dee01d767cefaa7/F3.jpg", "media/b4635dee01d767cefaa7/F4.jpg", "media/b4635dee01d767cefaa7/F5.jpg", "media/b4635dee01d767cefaa7/F6.jpg", "media/b4635dee01d767cefaa7/F7.jpg", "media/b4635dee01d767cefaa7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (white rally car with orange number 91) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. down\nC. up\nD. right\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e411979f927340b5037971f087d0a5add4e22a91258450c06ad1a3e9c42c03f2"}
{"qa_id": "youtube_vis__0b5b5e8e5a__camera_translation_direction__00", "video_id": "youtube_vis__0b5b5e8e5a", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "down", "up"], "frame_paths": ["media/e066a46bab00bb1885b2/F0.jpg", "media/e066a46bab00bb1885b2/F1.jpg", "media/e066a46bab00bb1885b2/F2.jpg", "media/e066a46bab00bb1885b2/F3.jpg", "media/e066a46bab00bb1885b2/F4.jpg", "media/e066a46bab00bb1885b2/F5.jpg", "media/e066a46bab00bb1885b2/F6.jpg", "media/e066a46bab00bb1885b2/F7.jpg", "media/e066a46bab00bb1885b2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3a9f6051a8ad3915ddf7ca2b77f9708e2a8e996a165edd6e51a5eafe07d00bca"}
{"qa_id": "youtube_vis__0b5b5e8e5a__object_world_motion_direction__01", "video_id": "youtube_vis__0b5b5e8e5a", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (person standing on the road) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "backward", "up", "right"], "frame_paths": ["media/8793514bf103cdb25d55/F0.jpg", "media/8793514bf103cdb25d55/F1.jpg", "media/8793514bf103cdb25d55/F2.jpg", "media/8793514bf103cdb25d55/F3.jpg", "media/8793514bf103cdb25d55/F4.jpg", "media/8793514bf103cdb25d55/F5.jpg", "media/8793514bf103cdb25d55/F6.jpg", "media/8793514bf103cdb25d55/F7.jpg", "media/8793514bf103cdb25d55/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (person standing on the road) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "07122953d9d0e4885266689dcd447dc5564a7d98ad8daf08e2d40d7967fa7da3"}
{"qa_id": "youtube_vis__0b7dbfa3cb__object_world_motion_direction__00", "video_id": "youtube_vis__0b7dbfa3cb", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow with white markings on the lower legs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "left", "backward", "up"], "frame_paths": ["media/a1346bff15be717c921e/F0.jpg", "media/a1346bff15be717c921e/F1.jpg", "media/a1346bff15be717c921e/F2.jpg", "media/a1346bff15be717c921e/F3.jpg", "media/a1346bff15be717c921e/F4.jpg", "media/a1346bff15be717c921e/F5.jpg", "media/a1346bff15be717c921e/F6.jpg", "media/a1346bff15be717c921e/F7.jpg", "media/a1346bff15be717c921e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow with white markings on the lower legs) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. left\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6bbe9a473b465dcc9f8f5cfe7e8a3639c5dcea6230a0b0a4173d23c23f50f33c"}
{"qa_id": "youtube_vis__0f17fa6fcb__object_world_motion_direction__00", "video_id": "youtube_vis__0f17fa6fcb", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F7, in which dominant direction did object B (mallard duck with green head and grey body) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "left", "backward", "forward"], "frame_paths": ["media/ff86b0483d63bf27d087/F0.jpg", "media/ff86b0483d63bf27d087/F1.jpg", "media/ff86b0483d63bf27d087/F2.jpg", "media/ff86b0483d63bf27d087/F3.jpg", "media/ff86b0483d63bf27d087/F4.jpg", "media/ff86b0483d63bf27d087/F5.jpg", "media/ff86b0483d63bf27d087/F6.jpg", "media/ff86b0483d63bf27d087/F7.jpg", "media/ff86b0483d63bf27d087/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F7, in which dominant direction did object B (mallard duck with green head and grey body) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4bff8d7702ca5b1b7355e43daa31bd34eedcf46b684b784a80c677ddb9d6ab9f"}
{"qa_id": "youtube_vis__0f17fa6fcb__object_pair_distance_change__01", "video_id": "youtube_vis__0f17fa6fcb", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, did the 3D distance between object A (mallard duck with orange head and white neck ring) and object B (mallard duck with green head and grey body) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/a4b3eab2f1ff12667259/F0.jpg", "media/a4b3eab2f1ff12667259/F1.jpg", "media/a4b3eab2f1ff12667259/F2.jpg", "media/a4b3eab2f1ff12667259/F3.jpg", "media/a4b3eab2f1ff12667259/F4.jpg", "media/a4b3eab2f1ff12667259/F5.jpg", "media/a4b3eab2f1ff12667259/F6.jpg", "media/a4b3eab2f1ff12667259/F7.jpg", "media/a4b3eab2f1ff12667259/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, did the 3D distance between object A (mallard duck with orange head and white neck ring) and object B (mallard duck with green head and grey body) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "709341444c689f6140b4e84020096ef20e76bcb09f1e7b42af9285cddf4da114"}
{"qa_id": "youtube_vis__1121b42b2e__object_world_motion_direction__00", "video_id": "youtube_vis__1121b42b2e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in blue shirt and black pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "forward", "up", "backward", "left"], "frame_paths": ["media/90ba530890544ec0e99d/F0.jpg", "media/90ba530890544ec0e99d/F1.jpg", "media/90ba530890544ec0e99d/F2.jpg", "media/90ba530890544ec0e99d/F3.jpg", "media/90ba530890544ec0e99d/F4.jpg", "media/90ba530890544ec0e99d/F5.jpg", "media/90ba530890544ec0e99d/F6.jpg", "media/90ba530890544ec0e99d/F7.jpg", "media/90ba530890544ec0e99d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (person in blue shirt and black pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. forward\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ef37dadc8ac4d895bec1324a9f46811cb1ded69dfaf49c4f09616a55f0510558"}
{"qa_id": "youtube_vis__1121b42b2e__camera_object_distance_change__01", "video_id": "youtube_vis__1121b42b2e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (person in blue shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/90ba530890544ec0e99d/F0.jpg", "media/90ba530890544ec0e99d/F1.jpg", "media/90ba530890544ec0e99d/F2.jpg", "media/90ba530890544ec0e99d/F3.jpg", "media/90ba530890544ec0e99d/F4.jpg", "media/90ba530890544ec0e99d/F5.jpg", "media/90ba530890544ec0e99d/F6.jpg", "media/90ba530890544ec0e99d/F7.jpg", "media/90ba530890544ec0e99d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (person in blue shirt and black pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b958e325dea3412409e3be6db4f3d6ce0bf59557695b6ffff59cbdbf587ffbc6"}
{"qa_id": "youtube_vis__122896672d__camera_view_direction_change__00", "video_id": "youtube_vis__122896672d", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/d7321d0285443a708a45/F0.jpg", "media/d7321d0285443a708a45/F1.jpg", "media/d7321d0285443a708a45/F2.jpg", "media/d7321d0285443a708a45/F3.jpg", "media/d7321d0285443a708a45/F4.jpg", "media/d7321d0285443a708a45/F5.jpg", "media/d7321d0285443a708a45/F6.jpg", "media/d7321d0285443a708a45/F7.jpg", "media/d7321d0285443a708a45/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "76bc240637684b4134ed087cc3ef3fc06616f0e520509034bda289d4710a18d7"}
{"qa_id": "youtube_vis__122896672d__camera_object_distance_change__01", "video_id": "youtube_vis__122896672d", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (person in light shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/789fb7b3b63a5b1c274c/F0.jpg", "media/789fb7b3b63a5b1c274c/F1.jpg", "media/789fb7b3b63a5b1c274c/F2.jpg", "media/789fb7b3b63a5b1c274c/F3.jpg", "media/789fb7b3b63a5b1c274c/F4.jpg", "media/789fb7b3b63a5b1c274c/F5.jpg", "media/789fb7b3b63a5b1c274c/F6.jpg", "media/789fb7b3b63a5b1c274c/F7.jpg", "media/789fb7b3b63a5b1c274c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in light shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a1436f2b5fbaeaee7530815fa3ef0e0c8928b507df9f741557ff8d80ed754563"}
{"qa_id": "youtube_vis__126d203967__camera_translation_direction__00", "video_id": "youtube_vis__126d203967", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "right", "forward", "down"], "frame_paths": ["media/a99b5f7a0b6c84591ca0/F0.jpg", "media/a99b5f7a0b6c84591ca0/F1.jpg", "media/a99b5f7a0b6c84591ca0/F2.jpg", "media/a99b5f7a0b6c84591ca0/F3.jpg", "media/a99b5f7a0b6c84591ca0/F4.jpg", "media/a99b5f7a0b6c84591ca0/F5.jpg", "media/a99b5f7a0b6c84591ca0/F6.jpg", "media/a99b5f7a0b6c84591ca0/F7.jpg", "media/a99b5f7a0b6c84591ca0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c422c4fc7aa9c5713ae0aac7f94e9bc56bdcab011e1a6eb88b6fea35b26e132a"}
{"qa_id": "youtube_vis__126d203967__object_world_motion_direction__01", "video_id": "youtube_vis__126d203967", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in red and blue striped shirt and helmet riding a skateboard) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "right", "left", "backward", "up", "forward"], "frame_paths": ["media/9505c13475c12b1024d3/F0.jpg", "media/9505c13475c12b1024d3/F1.jpg", "media/9505c13475c12b1024d3/F2.jpg", "media/9505c13475c12b1024d3/F3.jpg", "media/9505c13475c12b1024d3/F4.jpg", "media/9505c13475c12b1024d3/F5.jpg", "media/9505c13475c12b1024d3/F6.jpg", "media/9505c13475c12b1024d3/F7.jpg", "media/9505c13475c12b1024d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in red and blue striped shirt and helmet riding a skateboard) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. left\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8513dc64ee3d74e3ba78f98b94f9ef949985baf9ef8c08144fa3027cac6a76fc"}
{"qa_id": "youtube_vis__13e3070469__object_world_motion_direction__00", "video_id": "youtube_vis__13e3070469", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (adult zebra grazing in a grassy field) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "up", "down", "forward"], "frame_paths": ["media/79f5fa43feef1a84c944/F0.jpg", "media/79f5fa43feef1a84c944/F1.jpg", "media/79f5fa43feef1a84c944/F2.jpg", "media/79f5fa43feef1a84c944/F3.jpg", "media/79f5fa43feef1a84c944/F4.jpg", "media/79f5fa43feef1a84c944/F5.jpg", "media/79f5fa43feef1a84c944/F6.jpg", "media/79f5fa43feef1a84c944/F7.jpg", "media/79f5fa43feef1a84c944/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (adult zebra grazing in a grassy field) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. up\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "90236c6432619959bf1086f583793d6f82bd2c9411aafe0a2b3bd3273d8cea10"}
{"qa_id": "youtube_vis__13e3070469__object_pair_distance_change__01", "video_id": "youtube_vis__13e3070469", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did the 3D distance between object A (adult zebra grazing in a grassy field) and object B (zebra grazing on the left side of the frame) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/1f3d5117a0ecfa938f4b/F0.jpg", "media/1f3d5117a0ecfa938f4b/F1.jpg", "media/1f3d5117a0ecfa938f4b/F2.jpg", "media/1f3d5117a0ecfa938f4b/F3.jpg", "media/1f3d5117a0ecfa938f4b/F4.jpg", "media/1f3d5117a0ecfa938f4b/F5.jpg", "media/1f3d5117a0ecfa938f4b/F6.jpg", "media/1f3d5117a0ecfa938f4b/F7.jpg", "media/1f3d5117a0ecfa938f4b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (adult zebra grazing in a grassy field) and object B (zebra grazing on the left side of the frame) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87d087a12fb030e45fdbab5e13d199177f90776bc647a590af0faa069cc8341b"}
{"qa_id": "youtube_vis__145975e717__object_world_motion_direction__00", "video_id": "youtube_vis__145975e717", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in dark shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "backward", "left", "up", "down"], "frame_paths": ["media/65e7b849f084f51f2278/F0.jpg", "media/65e7b849f084f51f2278/F1.jpg", "media/65e7b849f084f51f2278/F2.jpg", "media/65e7b849f084f51f2278/F3.jpg", "media/65e7b849f084f51f2278/F4.jpg", "media/65e7b849f084f51f2278/F5.jpg", "media/65e7b849f084f51f2278/F6.jpg", "media/65e7b849f084f51f2278/F7.jpg", "media/65e7b849f084f51f2278/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in dark shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. backward\nD. left\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d5cf62a8d97585094eb669528a7cc66dc40adefee36468a1e78afc7b12fbd1d7"}
{"qa_id": "youtube_vis__145975e717__object_pair_distance_change__01", "video_id": "youtube_vis__145975e717", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, did the 3D distance between object A (person in white shirt and dark shorts) and object B (person in dark shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/d8ceb65ba1bc46de0a87/F0.jpg", "media/d8ceb65ba1bc46de0a87/F1.jpg", "media/d8ceb65ba1bc46de0a87/F2.jpg", "media/d8ceb65ba1bc46de0a87/F3.jpg", "media/d8ceb65ba1bc46de0a87/F4.jpg", "media/d8ceb65ba1bc46de0a87/F5.jpg", "media/d8ceb65ba1bc46de0a87/F6.jpg", "media/d8ceb65ba1bc46de0a87/F7.jpg", "media/d8ceb65ba1bc46de0a87/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did the 3D distance between object A (person in white shirt and dark shorts) and object B (person in dark shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aed83b75af66767024b39e8524ea3b7577f921c989b49bccc1e4a371e55f6569"}
{"qa_id": "youtube_vis__16af445362__camera_view_direction_change__00", "video_id": "youtube_vis__16af445362", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/c16b07c6d31319022944/F0.jpg", "media/c16b07c6d31319022944/F1.jpg", "media/c16b07c6d31319022944/F2.jpg", "media/c16b07c6d31319022944/F3.jpg", "media/c16b07c6d31319022944/F4.jpg", "media/c16b07c6d31319022944/F5.jpg", "media/c16b07c6d31319022944/F6.jpg", "media/c16b07c6d31319022944/F7.jpg", "media/c16b07c6d31319022944/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7e926e0307a4c7a1eb5448365b3a4094571acfdd8a70ab3b12f9cb3a87f944a2"}
{"qa_id": "youtube_vis__16af445362__camera_object_distance_change__01", "video_id": "youtube_vis__16af445362", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (dark grey military jet with a white 'A' on the tail fin) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/66c3002ab8c20b95fedf/F0.jpg", "media/66c3002ab8c20b95fedf/F1.jpg", "media/66c3002ab8c20b95fedf/F2.jpg", "media/66c3002ab8c20b95fedf/F3.jpg", "media/66c3002ab8c20b95fedf/F4.jpg", "media/66c3002ab8c20b95fedf/F5.jpg", "media/66c3002ab8c20b95fedf/F6.jpg", "media/66c3002ab8c20b95fedf/F7.jpg", "media/66c3002ab8c20b95fedf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (dark grey military jet with a white 'A' on the tail fin) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b886669c80b74c8defd0077faecaca020b9dceae864bbe1ca4d8353ae91b685"}
{"qa_id": "youtube_vis__18d7382a94__camera_translation_direction__00", "video_id": "youtube_vis__18d7382a94", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "up", "right", "backward", "down"], "frame_paths": ["media/3b23e11b1c3883b38d60/F0.jpg", "media/3b23e11b1c3883b38d60/F1.jpg", "media/3b23e11b1c3883b38d60/F2.jpg", "media/3b23e11b1c3883b38d60/F3.jpg", "media/3b23e11b1c3883b38d60/F4.jpg", "media/3b23e11b1c3883b38d60/F5.jpg", "media/3b23e11b1c3883b38d60/F6.jpg", "media/3b23e11b1c3883b38d60/F7.jpg", "media/3b23e11b1c3883b38d60/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. up\nD. right\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9211f64c5746848536c7b1e340501233a6d9cacb024e3c792cd6e4d00a56d4e6"}
{"qa_id": "youtube_vis__18d7382a94__camera_object_distance_change__01", "video_id": "youtube_vis__18d7382a94", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, did object A (person in white and blue athletic uniform with a white cap) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/636dc1907d7b6e06a271/F0.jpg", "media/636dc1907d7b6e06a271/F1.jpg", "media/636dc1907d7b6e06a271/F2.jpg", "media/636dc1907d7b6e06a271/F3.jpg", "media/636dc1907d7b6e06a271/F4.jpg", "media/636dc1907d7b6e06a271/F5.jpg", "media/636dc1907d7b6e06a271/F6.jpg", "media/636dc1907d7b6e06a271/F7.jpg", "media/636dc1907d7b6e06a271/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (person in white and blue athletic uniform with a white cap) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "61f8c8cdc7b922dc44697bb04af1958727e3a58a475a8391fd6e1e8d959036f7"}
{"qa_id": "youtube_vis__19ee80dac6__object_world_motion_direction__00", "video_id": "youtube_vis__19ee80dac6", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in black wetsuit paddling on a surfboard) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "left", "down", "right", "forward"], "frame_paths": ["media/a8aca689a9da44ff0d97/F0.jpg", "media/a8aca689a9da44ff0d97/F1.jpg", "media/a8aca689a9da44ff0d97/F2.jpg", "media/a8aca689a9da44ff0d97/F3.jpg", "media/a8aca689a9da44ff0d97/F4.jpg", "media/a8aca689a9da44ff0d97/F5.jpg", "media/a8aca689a9da44ff0d97/F6.jpg", "media/a8aca689a9da44ff0d97/F7.jpg", "media/a8aca689a9da44ff0d97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in black wetsuit paddling on a surfboard) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. left\nD. down\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f4aeb812a90b10a36bf9a5a934e37253df666eea0ce7cec0e71a5f8c364e389d"}
{"qa_id": "youtube_vis__1f7b3de3bc__camera_view_direction_change__00", "video_id": "youtube_vis__1f7b3de3bc", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/18e8a51e6c1de333d807/F0.jpg", "media/18e8a51e6c1de333d807/F1.jpg", "media/18e8a51e6c1de333d807/F2.jpg", "media/18e8a51e6c1de333d807/F3.jpg", "media/18e8a51e6c1de333d807/F4.jpg", "media/18e8a51e6c1de333d807/F5.jpg", "media/18e8a51e6c1de333d807/F6.jpg", "media/18e8a51e6c1de333d807/F7.jpg", "media/18e8a51e6c1de333d807/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb821ef5ac91183e7c2873f4974564eb33a0896e76a359527a26534ed1c81dfe"}
{"qa_id": "youtube_vis__1f7b3de3bc__object_world_motion_direction__01", "video_id": "youtube_vis__1f7b3de3bc", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (black race car with white and orange flame decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/446beac982cbc38b4e26/F0.jpg", "media/446beac982cbc38b4e26/F1.jpg", "media/446beac982cbc38b4e26/F2.jpg", "media/446beac982cbc38b4e26/F3.jpg", "media/446beac982cbc38b4e26/F4.jpg", "media/446beac982cbc38b4e26/F5.jpg", "media/446beac982cbc38b4e26/F6.jpg", "media/446beac982cbc38b4e26/F7.jpg", "media/446beac982cbc38b4e26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (black race car with white and orange flame decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8d271256d170112b46d4a70507e448c3af8cbc5061457091a01b4bb03996345b"}
{"qa_id": "youtube_vis__20a3d90790__object_world_motion_direction__00", "video_id": "youtube_vis__20a3d90790", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (runner in maroon shirt and orange shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "up", "forward", "right", "down"], "frame_paths": ["media/a8ecf5b07fb5be26e98c/F0.jpg", "media/a8ecf5b07fb5be26e98c/F1.jpg", "media/a8ecf5b07fb5be26e98c/F2.jpg", "media/a8ecf5b07fb5be26e98c/F3.jpg", "media/a8ecf5b07fb5be26e98c/F4.jpg", "media/a8ecf5b07fb5be26e98c/F5.jpg", "media/a8ecf5b07fb5be26e98c/F6.jpg", "media/a8ecf5b07fb5be26e98c/F7.jpg", "media/a8ecf5b07fb5be26e98c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (runner in maroon shirt and orange shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. up\nD. forward\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "74d94192785c3ea9ed5f5af7f2bbf55aba076b076c86b424c97a2a9453cbe4fc"}
{"qa_id": "youtube_vis__20a3d90790__object_world_motion_direction__01", "video_id": "youtube_vis__20a3d90790", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (runner in yellow shirt and black shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["left", "up", "forward", "backward", "down", "right"], "frame_paths": ["media/d0afffd5470bfa2a3610/F0.jpg", "media/d0afffd5470bfa2a3610/F1.jpg", "media/d0afffd5470bfa2a3610/F2.jpg", "media/d0afffd5470bfa2a3610/F3.jpg", "media/d0afffd5470bfa2a3610/F4.jpg", "media/d0afffd5470bfa2a3610/F5.jpg", "media/d0afffd5470bfa2a3610/F6.jpg", "media/d0afffd5470bfa2a3610/F7.jpg", "media/d0afffd5470bfa2a3610/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (runner in yellow shirt and black shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. forward\nD. backward\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f85b3b13c146526a1cdce60b2daba751b9032f183b627ee69bf712beadf4292a"}
{"qa_id": "youtube_vis__25d2163904__camera_object_distance_change__00", "video_id": "youtube_vis__25d2163904", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did object A (white race car with Prince Alibante livery) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b77e385e0cf13e005aa8/F0.jpg", "media/b77e385e0cf13e005aa8/F1.jpg", "media/b77e385e0cf13e005aa8/F2.jpg", "media/b77e385e0cf13e005aa8/F3.jpg", "media/b77e385e0cf13e005aa8/F4.jpg", "media/b77e385e0cf13e005aa8/F5.jpg", "media/b77e385e0cf13e005aa8/F6.jpg", "media/b77e385e0cf13e005aa8/F7.jpg", "media/b77e385e0cf13e005aa8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (white race car with Prince Alibante livery) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9b6f008aba920c113eddff0f18757508ef58af0c48bb4535e75cfc0ded2870d0"}
{"qa_id": "youtube_vis__25d2163904__object_pair_distance_change__01", "video_id": "youtube_vis__25d2163904", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F6, did the 3D distance between object A (white race car with Prince Alibante livery) and object B (man in black t-shirt and green cap) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/2852f83bfffef5ae76a9/F0.jpg", "media/2852f83bfffef5ae76a9/F1.jpg", "media/2852f83bfffef5ae76a9/F2.jpg", "media/2852f83bfffef5ae76a9/F3.jpg", "media/2852f83bfffef5ae76a9/F4.jpg", "media/2852f83bfffef5ae76a9/F5.jpg", "media/2852f83bfffef5ae76a9/F6.jpg", "media/2852f83bfffef5ae76a9/F7.jpg", "media/2852f83bfffef5ae76a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did the 3D distance between object A (white race car with Prince Alibante livery) and object B (man in black t-shirt and green cap) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5a83871b1d096464dd0b09c2824d7baba02f85d815ca557f1d42ac98431fc41"}
{"qa_id": "youtube_vis__25d2163904__object_travel_distance_comparison__02", "video_id": "youtube_vis__25d2163904", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white race car with Prince Alibante livery) or B (man in black t-shirt and green cap)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/2852f83bfffef5ae76a9/F0.jpg", "media/2852f83bfffef5ae76a9/F1.jpg", "media/2852f83bfffef5ae76a9/F2.jpg", "media/2852f83bfffef5ae76a9/F3.jpg", "media/2852f83bfffef5ae76a9/F4.jpg", "media/2852f83bfffef5ae76a9/F5.jpg", "media/2852f83bfffef5ae76a9/F6.jpg", "media/2852f83bfffef5ae76a9/F7.jpg", "media/2852f83bfffef5ae76a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F4 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white race car with Prince Alibante livery) or B (man in black t-shirt and green cap)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "25363a78c6c23b36471b80a87fb47a0484c669824e346a9b14c1f6ed0ba5168d"}
{"qa_id": "youtube_vis__2a821394e3__object_world_motion_direction__00", "video_id": "youtube_vis__2a821394e3", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (white and orange commuter train with bright headlights) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "left", "up", "down", "forward"], "frame_paths": ["media/22773d3ffb33e9fe76a8/F0.jpg", "media/22773d3ffb33e9fe76a8/F1.jpg", "media/22773d3ffb33e9fe76a8/F2.jpg", "media/22773d3ffb33e9fe76a8/F3.jpg", "media/22773d3ffb33e9fe76a8/F4.jpg", "media/22773d3ffb33e9fe76a8/F5.jpg", "media/22773d3ffb33e9fe76a8/F6.jpg", "media/22773d3ffb33e9fe76a8/F7.jpg", "media/22773d3ffb33e9fe76a8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (white and orange commuter train with bright headlights) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. left\nD. up\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9cbaeff39a48e15d7ac864fe23c7b474f4bb0bb38801785d1d74e739b8415ff6"}
{"qa_id": "youtube_vis__2ace57333e__camera_view_direction_change__00", "video_id": "youtube_vis__2ace57333e", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["down", "left", "right", "up"], "frame_paths": ["media/4b6350ebfa28c2cdf18e/F0.jpg", "media/4b6350ebfa28c2cdf18e/F1.jpg", "media/4b6350ebfa28c2cdf18e/F2.jpg", "media/4b6350ebfa28c2cdf18e/F3.jpg", "media/4b6350ebfa28c2cdf18e/F4.jpg", "media/4b6350ebfa28c2cdf18e/F5.jpg", "media/4b6350ebfa28c2cdf18e/F6.jpg", "media/4b6350ebfa28c2cdf18e/F7.jpg", "media/4b6350ebfa28c2cdf18e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. down\nB. left\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "55d5d0b42a861f65a9239fafb681083dab3aa12e12aae08c19aadab8d0059bc3"}
{"qa_id": "youtube_vis__2b53828ab5__object_world_motion_direction__00", "video_id": "youtube_vis__2b53828ab5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in white shirt and orange cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "down", "left", "up", "right"], "frame_paths": ["media/1d49b2f060d2b851eb9f/F0.jpg", "media/1d49b2f060d2b851eb9f/F1.jpg", "media/1d49b2f060d2b851eb9f/F2.jpg", "media/1d49b2f060d2b851eb9f/F3.jpg", "media/1d49b2f060d2b851eb9f/F4.jpg", "media/1d49b2f060d2b851eb9f/F5.jpg", "media/1d49b2f060d2b851eb9f/F6.jpg", "media/1d49b2f060d2b851eb9f/F7.jpg", "media/1d49b2f060d2b851eb9f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person in white shirt and orange cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. down\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7285bb50f7ca2e916b2cb9aa536c35d6d91ef304f10d5f4c062c0a40d188dcf5"}
{"qa_id": "youtube_vis__2b53828ab5__camera_object_distance_change__01", "video_id": "youtube_vis__2b53828ab5", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object A (person in white shirt and orange cap) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/1d49b2f060d2b851eb9f/F0.jpg", "media/1d49b2f060d2b851eb9f/F1.jpg", "media/1d49b2f060d2b851eb9f/F2.jpg", "media/1d49b2f060d2b851eb9f/F3.jpg", "media/1d49b2f060d2b851eb9f/F4.jpg", "media/1d49b2f060d2b851eb9f/F5.jpg", "media/1d49b2f060d2b851eb9f/F6.jpg", "media/1d49b2f060d2b851eb9f/F7.jpg", "media/1d49b2f060d2b851eb9f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person in white shirt and orange cap) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2126a3ab8783d081a1a7371943eb2cedb933e42a6339bac02834edfe50c1f296"}
{"qa_id": "youtube_vis__2b53828ab5__object_world_motion_direction__02", "video_id": "youtube_vis__2b53828ab5", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in red shirt and green cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "right", "forward", "backward", "up"], "frame_paths": ["media/28357b3243c7ff60b710/F0.jpg", "media/28357b3243c7ff60b710/F1.jpg", "media/28357b3243c7ff60b710/F2.jpg", "media/28357b3243c7ff60b710/F3.jpg", "media/28357b3243c7ff60b710/F4.jpg", "media/28357b3243c7ff60b710/F5.jpg", "media/28357b3243c7ff60b710/F6.jpg", "media/28357b3243c7ff60b710/F7.jpg", "media/28357b3243c7ff60b710/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (person in red shirt and green cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. right\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0695256fc370e13f95e993d3371fe22e663117c999685a28355b9b77cc805978"}
{"qa_id": "youtube_vis__2b53828ab5__camera_object_distance_change__03", "video_id": "youtube_vis__2b53828ab5", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object B (person in red shirt and green cap) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/28357b3243c7ff60b710/F0.jpg", "media/28357b3243c7ff60b710/F1.jpg", "media/28357b3243c7ff60b710/F2.jpg", "media/28357b3243c7ff60b710/F3.jpg", "media/28357b3243c7ff60b710/F4.jpg", "media/28357b3243c7ff60b710/F5.jpg", "media/28357b3243c7ff60b710/F6.jpg", "media/28357b3243c7ff60b710/F7.jpg", "media/28357b3243c7ff60b710/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (person in red shirt and green cap) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7385e50d188c731033f4a74807416f4d6379bc9cab5e2ab9587cd08a3149ed8c"}
{"qa_id": "youtube_vis__30d00e92ee__camera_view_direction_change__00", "video_id": "youtube_vis__30d00e92ee", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/67726269066882b25eac/F0.jpg", "media/67726269066882b25eac/F1.jpg", "media/67726269066882b25eac/F2.jpg", "media/67726269066882b25eac/F3.jpg", "media/67726269066882b25eac/F4.jpg", "media/67726269066882b25eac/F5.jpg", "media/67726269066882b25eac/F6.jpg", "media/67726269066882b25eac/F7.jpg", "media/67726269066882b25eac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f940fef843d2c19cca1df7c8a9d787c14654198ca1d67efcb885951d30953e8"}
{"qa_id": "youtube_vis__321b438662__camera_view_direction_change__00", "video_id": "youtube_vis__321b438662", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/325d48da826da286a83d/F0.jpg", "media/325d48da826da286a83d/F1.jpg", "media/325d48da826da286a83d/F2.jpg", "media/325d48da826da286a83d/F3.jpg", "media/325d48da826da286a83d/F4.jpg", "media/325d48da826da286a83d/F5.jpg", "media/325d48da826da286a83d/F6.jpg", "media/325d48da826da286a83d/F7.jpg", "media/325d48da826da286a83d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8cff1cdc4b54c5a8f11cad987f681201aadb2e007a2963b011b788db766fa114"}
{"qa_id": "youtube_vis__322f7bed9c__object_world_motion_direction__00", "video_id": "youtube_vis__322f7bed9c", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (skier in dark clothing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "forward", "left"], "frame_paths": ["media/83002aa6a33315fc78df/F0.jpg", "media/83002aa6a33315fc78df/F1.jpg", "media/83002aa6a33315fc78df/F2.jpg", "media/83002aa6a33315fc78df/F3.jpg", "media/83002aa6a33315fc78df/F4.jpg", "media/83002aa6a33315fc78df/F5.jpg", "media/83002aa6a33315fc78df/F6.jpg", "media/83002aa6a33315fc78df/F7.jpg", "media/83002aa6a33315fc78df/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (skier in dark clothing) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "36bb09134f95da243abaad40f8ca7c61b6480154e4f412161f33a3c38e44180d"}
{"qa_id": "youtube_vis__334cb835ac__camera_translation_direction__00", "video_id": "youtube_vis__334cb835ac", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "down", "right"], "frame_paths": ["media/1e99c57c153148e71753/F0.jpg", "media/1e99c57c153148e71753/F1.jpg", "media/1e99c57c153148e71753/F2.jpg", "media/1e99c57c153148e71753/F3.jpg", "media/1e99c57c153148e71753/F4.jpg", "media/1e99c57c153148e71753/F5.jpg", "media/1e99c57c153148e71753/F6.jpg", "media/1e99c57c153148e71753/F7.jpg", "media/1e99c57c153148e71753/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c7b95dc5ca37a76d859b5650475a3c444a14264aae166bb141f87f90b7ee4988"}
{"qa_id": "youtube_vis__334cb835ac__camera_view_direction_change__01", "video_id": "youtube_vis__334cb835ac", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/1e99c57c153148e71753/F0.jpg", "media/1e99c57c153148e71753/F1.jpg", "media/1e99c57c153148e71753/F2.jpg", "media/1e99c57c153148e71753/F3.jpg", "media/1e99c57c153148e71753/F4.jpg", "media/1e99c57c153148e71753/F5.jpg", "media/1e99c57c153148e71753/F6.jpg", "media/1e99c57c153148e71753/F7.jpg", "media/1e99c57c153148e71753/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F1?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eaa71c75bcc1a2b92934d26b46255be453ff117b49ba741cc880e65fcf158b7b"}
{"qa_id": "youtube_vis__334cb835ac__object_world_motion_direction__02", "video_id": "youtube_vis__334cb835ac", "task": "object_world_motion_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (snowboarder in dark clothing) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/3d9424ab72476bf0d67d/F0.jpg", "media/3d9424ab72476bf0d67d/F1.jpg", "media/3d9424ab72476bf0d67d/F2.jpg", "media/3d9424ab72476bf0d67d/F3.jpg", "media/3d9424ab72476bf0d67d/F4.jpg", "media/3d9424ab72476bf0d67d/F5.jpg", "media/3d9424ab72476bf0d67d/F6.jpg", "media/3d9424ab72476bf0d67d/F7.jpg", "media/3d9424ab72476bf0d67d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (snowboarder in dark clothing) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "83f1008cfde73b27d42bc4c508c583512dc66197b100717f99ac7ea44b5a138c"}
{"qa_id": "youtube_vis__334cb835ac__camera_object_distance_change__03", "video_id": "youtube_vis__334cb835ac", "task": "camera_object_distance_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F5, did object A (snowboarder in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/3d9424ab72476bf0d67d/F0.jpg", "media/3d9424ab72476bf0d67d/F1.jpg", "media/3d9424ab72476bf0d67d/F2.jpg", "media/3d9424ab72476bf0d67d/F3.jpg", "media/3d9424ab72476bf0d67d/F4.jpg", "media/3d9424ab72476bf0d67d/F5.jpg", "media/3d9424ab72476bf0d67d/F6.jpg", "media/3d9424ab72476bf0d67d/F7.jpg", "media/3d9424ab72476bf0d67d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (snowboarder in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4dd6c23b3d3106fcae06626aa2d6e1b4edb2face4428a357909c596d944522bd"}
{"qa_id": "youtube_vis__33a49e3ca4__camera_view_direction_change__00", "video_id": "youtube_vis__33a49e3ca4", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/ed18f4690182106b35fb/F0.jpg", "media/ed18f4690182106b35fb/F1.jpg", "media/ed18f4690182106b35fb/F2.jpg", "media/ed18f4690182106b35fb/F3.jpg", "media/ed18f4690182106b35fb/F4.jpg", "media/ed18f4690182106b35fb/F5.jpg", "media/ed18f4690182106b35fb/F6.jpg", "media/ed18f4690182106b35fb/F7.jpg", "media/ed18f4690182106b35fb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5bac693043594d1fe732e8d6be282a4d4e79d7e9c58acf4ef991c6952e1438e3"}
{"qa_id": "youtube_vis__3ae81609d6__camera_view_direction_change__00", "video_id": "youtube_vis__3ae81609d6", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/261000288f04a2c33db8/F0.jpg", "media/261000288f04a2c33db8/F1.jpg", "media/261000288f04a2c33db8/F2.jpg", "media/261000288f04a2c33db8/F3.jpg", "media/261000288f04a2c33db8/F4.jpg", "media/261000288f04a2c33db8/F5.jpg", "media/261000288f04a2c33db8/F6.jpg", "media/261000288f04a2c33db8/F7.jpg", "media/261000288f04a2c33db8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d3d3ce562c186879def57928d1d38a7173406c6c8fe51fd4dfa6050b8d18e39e"}
{"qa_id": "youtube_vis__3ae81609d6__object_world_motion_direction__01", "video_id": "youtube_vis__3ae81609d6", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (motorcyclist in orange jersey and helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "right", "forward", "left"], "frame_paths": ["media/b36d485507a90e7e2eb5/F0.jpg", "media/b36d485507a90e7e2eb5/F1.jpg", "media/b36d485507a90e7e2eb5/F2.jpg", "media/b36d485507a90e7e2eb5/F3.jpg", "media/b36d485507a90e7e2eb5/F4.jpg", "media/b36d485507a90e7e2eb5/F5.jpg", "media/b36d485507a90e7e2eb5/F6.jpg", "media/b36d485507a90e7e2eb5/F7.jpg", "media/b36d485507a90e7e2eb5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (motorcyclist in orange jersey and helmet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d409f85b3f40539c6055a48ca696595803340524bc1b5cc0c016cc56a72d1927"}
{"qa_id": "youtube_vis__3ae81609d6__camera_object_distance_change__02", "video_id": "youtube_vis__3ae81609d6", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (motorcyclist in orange jersey and helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b36d485507a90e7e2eb5/F0.jpg", "media/b36d485507a90e7e2eb5/F1.jpg", "media/b36d485507a90e7e2eb5/F2.jpg", "media/b36d485507a90e7e2eb5/F3.jpg", "media/b36d485507a90e7e2eb5/F4.jpg", "media/b36d485507a90e7e2eb5/F5.jpg", "media/b36d485507a90e7e2eb5/F6.jpg", "media/b36d485507a90e7e2eb5/F7.jpg", "media/b36d485507a90e7e2eb5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (motorcyclist in orange jersey and helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8ffdf5ed8adf13cd0a4122b58e820780bb32b0634f094375c63e69516924c170"}
{"qa_id": "youtube_vis__3bf4d2b823__camera_view_direction_change__00", "video_id": "youtube_vis__3bf4d2b823", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/72077ff2324cb9db1963/F0.jpg", "media/72077ff2324cb9db1963/F1.jpg", "media/72077ff2324cb9db1963/F2.jpg", "media/72077ff2324cb9db1963/F3.jpg", "media/72077ff2324cb9db1963/F4.jpg", "media/72077ff2324cb9db1963/F5.jpg", "media/72077ff2324cb9db1963/F6.jpg", "media/72077ff2324cb9db1963/F7.jpg", "media/72077ff2324cb9db1963/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "youtube_vis__3c6d549570__object_world_motion_direction__00", "video_id": "youtube_vis__3c6d549570", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (yellow car with black hood) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "right", "forward", "left"], "frame_paths": ["media/271b3997946c7319b44e/F0.jpg", "media/271b3997946c7319b44e/F1.jpg", "media/271b3997946c7319b44e/F2.jpg", "media/271b3997946c7319b44e/F3.jpg", "media/271b3997946c7319b44e/F4.jpg", "media/271b3997946c7319b44e/F5.jpg", "media/271b3997946c7319b44e/F6.jpg", "media/271b3997946c7319b44e/F7.jpg", "media/271b3997946c7319b44e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (yellow car with black hood) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fd08358ae6b752ef2917f7c694aaa96a21ec6946ddf69e00af63da562cab387b"}
{"qa_id": "youtube_vis__3c6d549570__camera_object_distance_change__01", "video_id": "youtube_vis__3c6d549570", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (yellow car with black hood) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/271b3997946c7319b44e/F0.jpg", "media/271b3997946c7319b44e/F1.jpg", "media/271b3997946c7319b44e/F2.jpg", "media/271b3997946c7319b44e/F3.jpg", "media/271b3997946c7319b44e/F4.jpg", "media/271b3997946c7319b44e/F5.jpg", "media/271b3997946c7319b44e/F6.jpg", "media/271b3997946c7319b44e/F7.jpg", "media/271b3997946c7319b44e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (yellow car with black hood) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0bf5156602ea3a5e2c057180f03b5fa7d31ce16e0e244d7cb1b8e263afef1445"}
{"qa_id": "youtube_vis__3c713cbf2f__object_world_motion_direction__00", "video_id": "youtube_vis__3c713cbf2f", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (white car drifting on a dirt track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/48427a9b9c8244406a46/F0.jpg", "media/48427a9b9c8244406a46/F1.jpg", "media/48427a9b9c8244406a46/F2.jpg", "media/48427a9b9c8244406a46/F3.jpg", "media/48427a9b9c8244406a46/F4.jpg", "media/48427a9b9c8244406a46/F5.jpg", "media/48427a9b9c8244406a46/F6.jpg", "media/48427a9b9c8244406a46/F7.jpg", "media/48427a9b9c8244406a46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (white car drifting on a dirt track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1a17b34dbfabe7a5092a27df2c6a29ce7aa966a516b3653f3dc8737808a4590b"}
{"qa_id": "youtube_vis__3c713cbf2f__camera_object_distance_change__01", "video_id": "youtube_vis__3c713cbf2f", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (white car drifting on a dirt track) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/48427a9b9c8244406a46/F0.jpg", "media/48427a9b9c8244406a46/F1.jpg", "media/48427a9b9c8244406a46/F2.jpg", "media/48427a9b9c8244406a46/F3.jpg", "media/48427a9b9c8244406a46/F4.jpg", "media/48427a9b9c8244406a46/F5.jpg", "media/48427a9b9c8244406a46/F6.jpg", "media/48427a9b9c8244406a46/F7.jpg", "media/48427a9b9c8244406a46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (white car drifting on a dirt track) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e226eac0b7c68be44e0f0ddcf9e5e9609bb706930b90b981290e97bb2cca9aae"}
{"qa_id": "youtube_vis__3e2845307e__camera_translation_direction__00", "video_id": "youtube_vis__3e2845307e", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/63c131b4f70276ac010c/F0.jpg", "media/63c131b4f70276ac010c/F1.jpg", "media/63c131b4f70276ac010c/F2.jpg", "media/63c131b4f70276ac010c/F3.jpg", "media/63c131b4f70276ac010c/F4.jpg", "media/63c131b4f70276ac010c/F5.jpg", "media/63c131b4f70276ac010c/F6.jpg", "media/63c131b4f70276ac010c/F7.jpg", "media/63c131b4f70276ac010c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c3c457add12e26327fface8aa64c1022fee9cf5af94bfeeec3ea2a2ff38e8318"}
{"qa_id": "youtube_vis__3e2845307e__object_world_motion_direction__01", "video_id": "youtube_vis__3e2845307e", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (large sea turtle swimming near the seabed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "right", "forward", "down"], "frame_paths": ["media/a264b24fc804b9c0e218/F0.jpg", "media/a264b24fc804b9c0e218/F1.jpg", "media/a264b24fc804b9c0e218/F2.jpg", "media/a264b24fc804b9c0e218/F3.jpg", "media/a264b24fc804b9c0e218/F4.jpg", "media/a264b24fc804b9c0e218/F5.jpg", "media/a264b24fc804b9c0e218/F6.jpg", "media/a264b24fc804b9c0e218/F7.jpg", "media/a264b24fc804b9c0e218/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (large sea turtle swimming near the seabed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d92dff727e2de3c54af45102cdaee57d0fb1867b738c94c093f1c1794ebdcf32"}
{"qa_id": "youtube_vis__40514eea6c__camera_view_direction_change__00", "video_id": "youtube_vis__40514eea6c", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/1f7e065449a991806c70/F0.jpg", "media/1f7e065449a991806c70/F1.jpg", "media/1f7e065449a991806c70/F2.jpg", "media/1f7e065449a991806c70/F3.jpg", "media/1f7e065449a991806c70/F4.jpg", "media/1f7e065449a991806c70/F5.jpg", "media/1f7e065449a991806c70/F6.jpg", "media/1f7e065449a991806c70/F7.jpg", "media/1f7e065449a991806c70/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "85b8fb8dd88fc3046f13517804383bd254fa352b40e4570a2ff8b991dce59d1f"}
{"qa_id": "youtube_vis__405874973b__object_world_motion_direction__00", "video_id": "youtube_vis__405874973b", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (large warthog with visible tusks and a greyish-brown coat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "right", "backward", "up", "forward"], "frame_paths": ["media/2ccade68f503dbd33a5f/F0.jpg", "media/2ccade68f503dbd33a5f/F1.jpg", "media/2ccade68f503dbd33a5f/F2.jpg", "media/2ccade68f503dbd33a5f/F3.jpg", "media/2ccade68f503dbd33a5f/F4.jpg", "media/2ccade68f503dbd33a5f/F5.jpg", "media/2ccade68f503dbd33a5f/F6.jpg", "media/2ccade68f503dbd33a5f/F7.jpg", "media/2ccade68f503dbd33a5f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (large warthog with visible tusks and a greyish-brown coat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. right\nD. backward\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3b6c896c8c44cd6e881c7e2613e8fc95e50c46ed7a34b5a7eca7b66bef929148"}
{"qa_id": "youtube_vis__41a34c20e7__object_world_motion_direction__00", "video_id": "youtube_vis__41a34c20e7", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and dark pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "right", "up", "left", "down"], "frame_paths": ["media/a23024cccb7a8d0778ab/F0.jpg", "media/a23024cccb7a8d0778ab/F1.jpg", "media/a23024cccb7a8d0778ab/F2.jpg", "media/a23024cccb7a8d0778ab/F3.jpg", "media/a23024cccb7a8d0778ab/F4.jpg", "media/a23024cccb7a8d0778ab/F5.jpg", "media/a23024cccb7a8d0778ab/F6.jpg", "media/a23024cccb7a8d0778ab/F7.jpg", "media/a23024cccb7a8d0778ab/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in blue shirt and dark pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. right\nD. up\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2c02233addc455af6904dd5b2efaf337788db374050b3cd341128dab5b988487"}
{"qa_id": "youtube_vis__429c3d0c93__camera_view_direction_change__00", "video_id": "youtube_vis__429c3d0c93", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/2ee7c34f83e544f0b33e/F0.jpg", "media/2ee7c34f83e544f0b33e/F1.jpg", "media/2ee7c34f83e544f0b33e/F2.jpg", "media/2ee7c34f83e544f0b33e/F3.jpg", "media/2ee7c34f83e544f0b33e/F4.jpg", "media/2ee7c34f83e544f0b33e/F5.jpg", "media/2ee7c34f83e544f0b33e/F6.jpg", "media/2ee7c34f83e544f0b33e/F7.jpg", "media/2ee7c34f83e544f0b33e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F5?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a87827a4ed7c9ce142cbe4605867f161c0b4b93743208d137581f47ecb0a7576"}
{"qa_id": "youtube_vis__429c3d0c93__object_world_motion_direction__01", "video_id": "youtube_vis__429c3d0c93", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (cyclist in black jersey and white helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "right", "left", "up"], "frame_paths": ["media/c2122fc7d7c8d383b2cf/F0.jpg", "media/c2122fc7d7c8d383b2cf/F1.jpg", "media/c2122fc7d7c8d383b2cf/F2.jpg", "media/c2122fc7d7c8d383b2cf/F3.jpg", "media/c2122fc7d7c8d383b2cf/F4.jpg", "media/c2122fc7d7c8d383b2cf/F5.jpg", "media/c2122fc7d7c8d383b2cf/F6.jpg", "media/c2122fc7d7c8d383b2cf/F7.jpg", "media/c2122fc7d7c8d383b2cf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (cyclist in black jersey and white helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "53f4d1ba9e3248c2fced8c90fcc9880ecd93f5fea396507b089813ec5b86d69d"}
{"qa_id": "youtube_vis__43d63b752a__object_world_motion_direction__00", "video_id": "youtube_vis__43d63b752a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (snowboarder in black jacket and green pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "down", "left", "right"], "frame_paths": ["media/e14d4d013989902e71a1/F0.jpg", "media/e14d4d013989902e71a1/F1.jpg", "media/e14d4d013989902e71a1/F2.jpg", "media/e14d4d013989902e71a1/F3.jpg", "media/e14d4d013989902e71a1/F4.jpg", "media/e14d4d013989902e71a1/F5.jpg", "media/e14d4d013989902e71a1/F6.jpg", "media/e14d4d013989902e71a1/F7.jpg", "media/e14d4d013989902e71a1/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (snowboarder in black jacket and green pants) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. down\nE. left\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fef0d56d2c5b677ba9b068b3883880de3c59ab4ab71998db1bb6d405f41e4f24"}
{"qa_id": "youtube_vis__483c72fb4d__object_world_motion_direction__00", "video_id": "youtube_vis__483c72fb4d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in white shirt and white shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/a5589b9105433e876896/F0.jpg", "media/a5589b9105433e876896/F1.jpg", "media/a5589b9105433e876896/F2.jpg", "media/a5589b9105433e876896/F3.jpg", "media/a5589b9105433e876896/F4.jpg", "media/a5589b9105433e876896/F5.jpg", "media/a5589b9105433e876896/F6.jpg", "media/a5589b9105433e876896/F7.jpg", "media/a5589b9105433e876896/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object A (person in white shirt and white shorts) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e1a866e520b141b3ca33e91df7e243b6ea5b915cc33bc307fdec2f9b29594308"}
{"qa_id": "youtube_vis__483c72fb4d__object_pair_distance_change__01", "video_id": "youtube_vis__483c72fb4d", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F5, did the 3D distance between object A (person in white shirt and white shorts) and object B (person in grey shirt and white shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/d31177366f257d13238d/F0.jpg", "media/d31177366f257d13238d/F1.jpg", "media/d31177366f257d13238d/F2.jpg", "media/d31177366f257d13238d/F3.jpg", "media/d31177366f257d13238d/F4.jpg", "media/d31177366f257d13238d/F5.jpg", "media/d31177366f257d13238d/F6.jpg", "media/d31177366f257d13238d/F7.jpg", "media/d31177366f257d13238d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, did the 3D distance between object A (person in white shirt and white shorts) and object B (person in grey shirt and white shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "10bd0f49900e9f97978afe9e20b9e6640aaec8ea338c60f34585fad711dc5ce2"}
{"qa_id": "youtube_vis__483c72fb4d__object_travel_distance_comparison__02", "video_id": "youtube_vis__483c72fb4d", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and white shorts) or B (person in grey shirt and white shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/d31177366f257d13238d/F0.jpg", "media/d31177366f257d13238d/F1.jpg", "media/d31177366f257d13238d/F2.jpg", "media/d31177366f257d13238d/F3.jpg", "media/d31177366f257d13238d/F4.jpg", "media/d31177366f257d13238d/F5.jpg", "media/d31177366f257d13238d/F6.jpg", "media/d31177366f257d13238d/F7.jpg", "media/d31177366f257d13238d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in white shirt and white shorts) or B (person in grey shirt and white shorts)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a4c3e9e0a87f4e0d832203eb67cb993238009114b91aa5cab43fe308dd07067"}
{"qa_id": "youtube_vis__4e3a9486f2__camera_translation_direction__00", "video_id": "youtube_vis__4e3a9486f2", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "up", "forward", "left"], "frame_paths": ["media/c397808b35b71f8f1530/F0.jpg", "media/c397808b35b71f8f1530/F1.jpg", "media/c397808b35b71f8f1530/F2.jpg", "media/c397808b35b71f8f1530/F3.jpg", "media/c397808b35b71f8f1530/F4.jpg", "media/c397808b35b71f8f1530/F5.jpg", "media/c397808b35b71f8f1530/F6.jpg", "media/c397808b35b71f8f1530/F7.jpg", "media/c397808b35b71f8f1530/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f47606addd5a5a6d02fdbd758dfcd828e4611ec66c815865e48bf4dcd332820f"}
{"qa_id": "youtube_vis__4eb6fc23a2__camera_translation_direction__00", "video_id": "youtube_vis__4eb6fc23a2", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "right", "up", "forward", "left", "backward"], "frame_paths": ["media/de4e21e6f4ea1a901e36/F0.jpg", "media/de4e21e6f4ea1a901e36/F1.jpg", "media/de4e21e6f4ea1a901e36/F2.jpg", "media/de4e21e6f4ea1a901e36/F3.jpg", "media/de4e21e6f4ea1a901e36/F4.jpg", "media/de4e21e6f4ea1a901e36/F5.jpg", "media/de4e21e6f4ea1a901e36/F6.jpg", "media/de4e21e6f4ea1a901e36/F7.jpg", "media/de4e21e6f4ea1a901e36/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. right\nC. up\nD. forward\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "027662033e6637c125cd25d1511e3dd7ed0c90007b8ced1c0dea3e1ca79a7ad1"}
{"qa_id": "youtube_vis__4fdfef4dea__camera_view_direction_change__00", "video_id": "youtube_vis__4fdfef4dea", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "left", "up", "right"], "frame_paths": ["media/aadad687acc93693fb3b/F0.jpg", "media/aadad687acc93693fb3b/F1.jpg", "media/aadad687acc93693fb3b/F2.jpg", "media/aadad687acc93693fb3b/F3.jpg", "media/aadad687acc93693fb3b/F4.jpg", "media/aadad687acc93693fb3b/F5.jpg", "media/aadad687acc93693fb3b/F6.jpg", "media/aadad687acc93693fb3b/F7.jpg", "media/aadad687acc93693fb3b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. left\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64e0fec494c815ab1788b70a6ff118230da5d701e043aa379365e44f1a46e68d"}
{"qa_id": "youtube_vis__4fdfef4dea__object_world_motion_direction__01", "video_id": "youtube_vis__4fdfef4dea", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object A (cyclist in white and blue jersey with orange helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "left", "right", "up"], "frame_paths": ["media/af1504e683307142d019/F0.jpg", "media/af1504e683307142d019/F1.jpg", "media/af1504e683307142d019/F2.jpg", "media/af1504e683307142d019/F3.jpg", "media/af1504e683307142d019/F4.jpg", "media/af1504e683307142d019/F5.jpg", "media/af1504e683307142d019/F6.jpg", "media/af1504e683307142d019/F7.jpg", "media/af1504e683307142d019/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object A (cyclist in white and blue jersey with orange helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. left\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "705933b3246be17ffb8e8dc9b246eff9659a7f4ae716cb1aa2744def0e7844a2"}
{"qa_id": "youtube_vis__4fdfef4dea__object_world_motion_direction__02", "video_id": "youtube_vis__4fdfef4dea", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F8, in which dominant direction did object B (cyclist in black and white jersey with green helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "left", "forward", "down"], "frame_paths": ["media/767c2d31437a1e6056a4/F0.jpg", "media/767c2d31437a1e6056a4/F1.jpg", "media/767c2d31437a1e6056a4/F2.jpg", "media/767c2d31437a1e6056a4/F3.jpg", "media/767c2d31437a1e6056a4/F4.jpg", "media/767c2d31437a1e6056a4/F5.jpg", "media/767c2d31437a1e6056a4/F6.jpg", "media/767c2d31437a1e6056a4/F7.jpg", "media/767c2d31437a1e6056a4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F8, in which dominant direction did object B (cyclist in black and white jersey with green helmet) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cdac02f7835f1b3022b06480de96446bf9f9b1bd6016976e1e73cc780a631c41"}
{"qa_id": "youtube_vis__4fdfef4dea__object_pair_distance_change__03", "video_id": "youtube_vis__4fdfef4dea", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did the 3D distance between object A (cyclist in white and blue jersey with orange helmet) and object B (cyclist in black and white jersey with green helmet) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/691c8806effc86a56189/F0.jpg", "media/691c8806effc86a56189/F1.jpg", "media/691c8806effc86a56189/F2.jpg", "media/691c8806effc86a56189/F3.jpg", "media/691c8806effc86a56189/F4.jpg", "media/691c8806effc86a56189/F5.jpg", "media/691c8806effc86a56189/F6.jpg", "media/691c8806effc86a56189/F7.jpg", "media/691c8806effc86a56189/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (cyclist in white and blue jersey with orange helmet) and object B (cyclist in black and white jersey with green helmet) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64f15312942e6958bc3f8365dae99658bdb4656859b034fd3e40b191df2ce873"}
{"qa_id": "youtube_vis__4fdfef4dea__object_travel_distance_comparison__04", "video_id": "youtube_vis__4fdfef4dea", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (cyclist in white and blue jersey with orange helmet) or B (cyclist in black and white jersey with green helmet)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/691c8806effc86a56189/F0.jpg", "media/691c8806effc86a56189/F1.jpg", "media/691c8806effc86a56189/F2.jpg", "media/691c8806effc86a56189/F3.jpg", "media/691c8806effc86a56189/F4.jpg", "media/691c8806effc86a56189/F5.jpg", "media/691c8806effc86a56189/F6.jpg", "media/691c8806effc86a56189/F7.jpg", "media/691c8806effc86a56189/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (cyclist in white and blue jersey with orange helmet) or B (cyclist in black and white jersey with green helmet)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b1f833c8158d5c09e891bfc44f436f5b4ee8d1a2f2bb080a81d28d5a1431f392"}
{"qa_id": "youtube_vis__4feb3ac01f__camera_translation_direction__00", "video_id": "youtube_vis__4feb3ac01f", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "backward", "down", "left"], "frame_paths": ["media/610d716e8db5088b90e6/F0.jpg", "media/610d716e8db5088b90e6/F1.jpg", "media/610d716e8db5088b90e6/F2.jpg", "media/610d716e8db5088b90e6/F3.jpg", "media/610d716e8db5088b90e6/F4.jpg", "media/610d716e8db5088b90e6/F5.jpg", "media/610d716e8db5088b90e6/F6.jpg", "media/610d716e8db5088b90e6/F7.jpg", "media/610d716e8db5088b90e6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6eed6963a6b88722b2e7db736827a3b6d9b1c04db2756965c650124431b67dc5"}
{"qa_id": "youtube_vis__55201f1d7e__camera_view_direction_change__00", "video_id": "youtube_vis__55201f1d7e", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "right", "up", "down"], "frame_paths": ["media/88e8439efeedf16cdb97/F0.jpg", "media/88e8439efeedf16cdb97/F1.jpg", "media/88e8439efeedf16cdb97/F2.jpg", "media/88e8439efeedf16cdb97/F3.jpg", "media/88e8439efeedf16cdb97/F4.jpg", "media/88e8439efeedf16cdb97/F5.jpg", "media/88e8439efeedf16cdb97/F6.jpg", "media/88e8439efeedf16cdb97/F7.jpg", "media/88e8439efeedf16cdb97/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. right\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2a3d38d4d49c976984b990e18699b523b069bef24a93e01b1277603400a3ffe7"}
{"qa_id": "youtube_vis__5883c6e381__camera_view_direction_change__00", "video_id": "youtube_vis__5883c6e381", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/d8c5d132fb1cb2ed7cef/F0.jpg", "media/d8c5d132fb1cb2ed7cef/F1.jpg", "media/d8c5d132fb1cb2ed7cef/F2.jpg", "media/d8c5d132fb1cb2ed7cef/F3.jpg", "media/d8c5d132fb1cb2ed7cef/F4.jpg", "media/d8c5d132fb1cb2ed7cef/F5.jpg", "media/d8c5d132fb1cb2ed7cef/F6.jpg", "media/d8c5d132fb1cb2ed7cef/F7.jpg", "media/d8c5d132fb1cb2ed7cef/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9557c0796e28ae2a4f32990ec2b747f1a5933b7979d70ae5cf47b924054e1483"}
{"qa_id": "youtube_vis__588f925ab6__object_world_motion_direction__00", "video_id": "youtube_vis__588f925ab6", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in black clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/dc71d6726e48a2804a01/F0.jpg", "media/dc71d6726e48a2804a01/F1.jpg", "media/dc71d6726e48a2804a01/F2.jpg", "media/dc71d6726e48a2804a01/F3.jpg", "media/dc71d6726e48a2804a01/F4.jpg", "media/dc71d6726e48a2804a01/F5.jpg", "media/dc71d6726e48a2804a01/F6.jpg", "media/dc71d6726e48a2804a01/F7.jpg", "media/dc71d6726e48a2804a01/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (person in black clothing) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e3955e1b0a890d81fabbe9c68907795d7817ec7d2b7ddbcf4bfd249cb830342b"}
{"qa_id": "youtube_vis__588f925ab6__camera_object_distance_change__01", "video_id": "youtube_vis__588f925ab6", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (person in black clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/dc71d6726e48a2804a01/F0.jpg", "media/dc71d6726e48a2804a01/F1.jpg", "media/dc71d6726e48a2804a01/F2.jpg", "media/dc71d6726e48a2804a01/F3.jpg", "media/dc71d6726e48a2804a01/F4.jpg", "media/dc71d6726e48a2804a01/F5.jpg", "media/dc71d6726e48a2804a01/F6.jpg", "media/dc71d6726e48a2804a01/F7.jpg", "media/dc71d6726e48a2804a01/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person in black clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "360203344960dbac08842a2a9e62c2271c5631f179618097a569b84d29198d33"}
{"qa_id": "youtube_vis__5892409fbd__camera_translation_direction__00", "video_id": "youtube_vis__5892409fbd", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "right", "left", "down", "backward"], "frame_paths": ["media/08968656dccf362be57f/F0.jpg", "media/08968656dccf362be57f/F1.jpg", "media/08968656dccf362be57f/F2.jpg", "media/08968656dccf362be57f/F3.jpg", "media/08968656dccf362be57f/F4.jpg", "media/08968656dccf362be57f/F5.jpg", "media/08968656dccf362be57f/F6.jpg", "media/08968656dccf362be57f/F7.jpg", "media/08968656dccf362be57f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "aa9acd363f8cb2a59b73436f4a7c8e9753585eefe7c335703daf0f7fdf150ffa"}
{"qa_id": "youtube_vis__5a50640995__camera_view_direction_change__00", "video_id": "youtube_vis__5a50640995", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "right", "down", "up"], "frame_paths": ["media/e469d17154304547fa26/F0.jpg", "media/e469d17154304547fa26/F1.jpg", "media/e469d17154304547fa26/F2.jpg", "media/e469d17154304547fa26/F3.jpg", "media/e469d17154304547fa26/F4.jpg", "media/e469d17154304547fa26/F5.jpg", "media/e469d17154304547fa26/F6.jpg", "media/e469d17154304547fa26/F7.jpg", "media/e469d17154304547fa26/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. right\nC. down\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a125cfc0f77d2ae9b2e6bc05b025870e8cceaf01ceee8bce113b53b93422203e"}
{"qa_id": "youtube_vis__5a50640995__object_world_motion_direction__01", "video_id": "youtube_vis__5a50640995", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (white goose with orange beak) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "down", "backward", "up", "left"], "frame_paths": ["media/43e1ee940cd1f045237c/F0.jpg", "media/43e1ee940cd1f045237c/F1.jpg", "media/43e1ee940cd1f045237c/F2.jpg", "media/43e1ee940cd1f045237c/F3.jpg", "media/43e1ee940cd1f045237c/F4.jpg", "media/43e1ee940cd1f045237c/F5.jpg", "media/43e1ee940cd1f045237c/F6.jpg", "media/43e1ee940cd1f045237c/F7.jpg", "media/43e1ee940cd1f045237c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3", "B": "g5"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (white goose with orange beak) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. down\nD. backward\nE. up\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "168914005eca71b09fc613f484cfb0ad43f1bd41d35e94ace88c0b1a65e1c64b"}
{"qa_id": "youtube_vis__5a5a1d6632__camera_translation_direction__00", "video_id": "youtube_vis__5a5a1d6632", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "left", "backward", "right", "down", "forward"], "frame_paths": ["media/caf6aabc20a87ff5e433/F0.jpg", "media/caf6aabc20a87ff5e433/F1.jpg", "media/caf6aabc20a87ff5e433/F2.jpg", "media/caf6aabc20a87ff5e433/F3.jpg", "media/caf6aabc20a87ff5e433/F4.jpg", "media/caf6aabc20a87ff5e433/F5.jpg", "media/caf6aabc20a87ff5e433/F6.jpg", "media/caf6aabc20a87ff5e433/F7.jpg", "media/caf6aabc20a87ff5e433/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. left\nC. backward\nD. right\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6af52e2796baa8a0b5a06a516ccb981433d41f3a8e33f23281ed2a0a826eb0e6"}
{"qa_id": "youtube_vis__5ce3e74cbc__camera_view_direction_change__00", "video_id": "youtube_vis__5ce3e74cbc", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/9d3ef440b4706490ce39/F0.jpg", "media/9d3ef440b4706490ce39/F1.jpg", "media/9d3ef440b4706490ce39/F2.jpg", "media/9d3ef440b4706490ce39/F3.jpg", "media/9d3ef440b4706490ce39/F4.jpg", "media/9d3ef440b4706490ce39/F5.jpg", "media/9d3ef440b4706490ce39/F6.jpg", "media/9d3ef440b4706490ce39/F7.jpg", "media/9d3ef440b4706490ce39/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "efdce93fb3432f829e04ecb89942991927c71ea8847b637914c4337a6151f339"}
{"qa_id": "youtube_vis__5ce3e74cbc__object_world_motion_direction__01", "video_id": "youtube_vis__5ce3e74cbc", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (female athlete in red tank top and shorts running) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "backward", "up", "right"], "frame_paths": ["media/08e9038c234f30575f3c/F0.jpg", "media/08e9038c234f30575f3c/F1.jpg", "media/08e9038c234f30575f3c/F2.jpg", "media/08e9038c234f30575f3c/F3.jpg", "media/08e9038c234f30575f3c/F4.jpg", "media/08e9038c234f30575f3c/F5.jpg", "media/08e9038c234f30575f3c/F6.jpg", "media/08e9038c234f30575f3c/F7.jpg", "media/08e9038c234f30575f3c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (female athlete in red tank top and shorts running) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "182411bdc1bac04ee86b7eb4798041a24954c3059fc6dba14e6a62bb16c8cd02"}
{"qa_id": "youtube_vis__5ce3e74cbc__camera_object_distance_change__02", "video_id": "youtube_vis__5ce3e74cbc", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did object A (female athlete in red tank top and shorts running) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/08e9038c234f30575f3c/F0.jpg", "media/08e9038c234f30575f3c/F1.jpg", "media/08e9038c234f30575f3c/F2.jpg", "media/08e9038c234f30575f3c/F3.jpg", "media/08e9038c234f30575f3c/F4.jpg", "media/08e9038c234f30575f3c/F5.jpg", "media/08e9038c234f30575f3c/F6.jpg", "media/08e9038c234f30575f3c/F7.jpg", "media/08e9038c234f30575f3c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (female athlete in red tank top and shorts running) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a44cda0a0e501c6ae47362df3375d5efe1837c36ad79ce6e66e48eab1f939a61"}
{"qa_id": "youtube_vis__5e8afccb05__camera_object_distance_change__00", "video_id": "youtube_vis__5e8afccb05", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object A (orange van) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/95898c4051d4daeeb8e0/F0.jpg", "media/95898c4051d4daeeb8e0/F1.jpg", "media/95898c4051d4daeeb8e0/F2.jpg", "media/95898c4051d4daeeb8e0/F3.jpg", "media/95898c4051d4daeeb8e0/F4.jpg", "media/95898c4051d4daeeb8e0/F5.jpg", "media/95898c4051d4daeeb8e0/F6.jpg", "media/95898c4051d4daeeb8e0/F7.jpg", "media/95898c4051d4daeeb8e0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object A (orange van) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f14880102ddfd22ea14e085a9e68b32a62d623095a43ff02c2ae046e9e73fdd5"}
{"qa_id": "youtube_vis__60c0873f42__camera_view_direction_change__00", "video_id": "youtube_vis__60c0873f42", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "left", "up"], "frame_paths": ["media/7a83dafe36008a9cd56f/F0.jpg", "media/7a83dafe36008a9cd56f/F1.jpg", "media/7a83dafe36008a9cd56f/F2.jpg", "media/7a83dafe36008a9cd56f/F3.jpg", "media/7a83dafe36008a9cd56f/F4.jpg", "media/7a83dafe36008a9cd56f/F5.jpg", "media/7a83dafe36008a9cd56f/F6.jpg", "media/7a83dafe36008a9cd56f/F7.jpg", "media/7a83dafe36008a9cd56f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c709c05b5e9f642838200dd1db01497476ff7fb5e69d584e04c52addbe602c91"}
{"qa_id": "youtube_vis__60c0873f42__camera_object_distance_change__01", "video_id": "youtube_vis__60c0873f42", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F2 to F4, did object A (person in white shirt and dark shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fb81710eba9b33fca3f9/F0.jpg", "media/fb81710eba9b33fca3f9/F1.jpg", "media/fb81710eba9b33fca3f9/F2.jpg", "media/fb81710eba9b33fca3f9/F3.jpg", "media/fb81710eba9b33fca3f9/F4.jpg", "media/fb81710eba9b33fca3f9/F5.jpg", "media/fb81710eba9b33fca3f9/F6.jpg", "media/fb81710eba9b33fca3f9/F7.jpg", "media/fb81710eba9b33fca3f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (person in white shirt and dark shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "776d70d4c2e5f339a108a35a2947fe8ad56a1834a0c8443591e54bde88e3d16b"}
{"qa_id": "youtube_vis__614e7078db__camera_view_direction_change__00", "video_id": "youtube_vis__614e7078db", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/6193c129b03149ca1cd9/F0.jpg", "media/6193c129b03149ca1cd9/F1.jpg", "media/6193c129b03149ca1cd9/F2.jpg", "media/6193c129b03149ca1cd9/F3.jpg", "media/6193c129b03149ca1cd9/F4.jpg", "media/6193c129b03149ca1cd9/F5.jpg", "media/6193c129b03149ca1cd9/F6.jpg", "media/6193c129b03149ca1cd9/F7.jpg", "media/6193c129b03149ca1cd9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cc01c8f565adc7d70c3dfe8415bb1beb47a7031980d38a04b07a94aa6f06829f"}
{"qa_id": "youtube_vis__614e7078db__camera_object_distance_change__01", "video_id": "youtube_vis__614e7078db", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (person wearing an orange life vest) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/432e83c926a4ee219c32/F0.jpg", "media/432e83c926a4ee219c32/F1.jpg", "media/432e83c926a4ee219c32/F2.jpg", "media/432e83c926a4ee219c32/F3.jpg", "media/432e83c926a4ee219c32/F4.jpg", "media/432e83c926a4ee219c32/F5.jpg", "media/432e83c926a4ee219c32/F6.jpg", "media/432e83c926a4ee219c32/F7.jpg", "media/432e83c926a4ee219c32/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person wearing an orange life vest) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e0c2ca881bf1d6d80465a711ba59a813b7b24d3937bea7d4413f405a551a9b24"}
{"qa_id": "youtube_vis__614e7078db__object_world_motion_direction__02", "video_id": "youtube_vis__614e7078db", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (person wearing a green life vest) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "right", "backward", "down", "left"], "frame_paths": ["media/7dcef3de13e2395ac41a/F0.jpg", "media/7dcef3de13e2395ac41a/F1.jpg", "media/7dcef3de13e2395ac41a/F2.jpg", "media/7dcef3de13e2395ac41a/F3.jpg", "media/7dcef3de13e2395ac41a/F4.jpg", "media/7dcef3de13e2395ac41a/F5.jpg", "media/7dcef3de13e2395ac41a/F6.jpg", "media/7dcef3de13e2395ac41a/F7.jpg", "media/7dcef3de13e2395ac41a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (person wearing a green life vest) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. right\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb4b7f911cd05dfc211b66bc092af40f7380da425f55ec1130c0472c6fa8b943"}
{"qa_id": "youtube_vis__614e7078db__camera_object_distance_change__03", "video_id": "youtube_vis__614e7078db", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (person wearing a green life vest) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/7dcef3de13e2395ac41a/F0.jpg", "media/7dcef3de13e2395ac41a/F1.jpg", "media/7dcef3de13e2395ac41a/F2.jpg", "media/7dcef3de13e2395ac41a/F3.jpg", "media/7dcef3de13e2395ac41a/F4.jpg", "media/7dcef3de13e2395ac41a/F5.jpg", "media/7dcef3de13e2395ac41a/F6.jpg", "media/7dcef3de13e2395ac41a/F7.jpg", "media/7dcef3de13e2395ac41a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (person wearing a green life vest) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a76a840b6fb0d1baba1d01790ec1b1268d471292b826b002270e14c1b18866d2"}
{"qa_id": "youtube_vis__62d6ece152__object_world_motion_direction__00", "video_id": "youtube_vis__62d6ece152", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F5, in which dominant direction did object B (light-colored electric guitar) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "right", "left", "up", "backward"], "frame_paths": ["media/bca0681f00272d4e4f1a/F0.jpg", "media/bca0681f00272d4e4f1a/F1.jpg", "media/bca0681f00272d4e4f1a/F2.jpg", "media/bca0681f00272d4e4f1a/F3.jpg", "media/bca0681f00272d4e4f1a/F4.jpg", "media/bca0681f00272d4e4f1a/F5.jpg", "media/bca0681f00272d4e4f1a/F6.jpg", "media/bca0681f00272d4e4f1a/F7.jpg", "media/bca0681f00272d4e4f1a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F5, in which dominant direction did object B (light-colored electric guitar) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. right\nD. left\nE. up\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f5bdda13e192d327daa3f4cd71d1c1f8aa8956b252cdd671dd368239f0ead5a4"}
{"qa_id": "youtube_vis__644081b88d__camera_translation_direction__00", "video_id": "youtube_vis__644081b88d", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/963183a4a0d700468023/F0.jpg", "media/963183a4a0d700468023/F1.jpg", "media/963183a4a0d700468023/F2.jpg", "media/963183a4a0d700468023/F3.jpg", "media/963183a4a0d700468023/F4.jpg", "media/963183a4a0d700468023/F5.jpg", "media/963183a4a0d700468023/F6.jpg", "media/963183a4a0d700468023/F7.jpg", "media/963183a4a0d700468023/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "89c2352bfb4846e576d14f4a6eccc20382a5593a762cd75de1ecc31324b4a34e"}
{"qa_id": "youtube_vis__644081b88d__object_world_motion_direction__01", "video_id": "youtube_vis__644081b88d", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (dark fish with a long, thin tail swimming near the seabed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "right", "forward", "left"], "frame_paths": ["media/7b910f736c81640b99da/F0.jpg", "media/7b910f736c81640b99da/F1.jpg", "media/7b910f736c81640b99da/F2.jpg", "media/7b910f736c81640b99da/F3.jpg", "media/7b910f736c81640b99da/F4.jpg", "media/7b910f736c81640b99da/F5.jpg", "media/7b910f736c81640b99da/F6.jpg", "media/7b910f736c81640b99da/F7.jpg", "media/7b910f736c81640b99da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (dark fish with a long, thin tail swimming near the seabed) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cf806e1f37d42f06d4ef4b89a5da43241216b174c7f358f18b135f659a228cd4"}
{"qa_id": "youtube_vis__6797f20aa9__camera_view_direction_change__00", "video_id": "youtube_vis__6797f20aa9", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/2096f52618a3171db421/F0.jpg", "media/2096f52618a3171db421/F1.jpg", "media/2096f52618a3171db421/F2.jpg", "media/2096f52618a3171db421/F3.jpg", "media/2096f52618a3171db421/F4.jpg", "media/2096f52618a3171db421/F5.jpg", "media/2096f52618a3171db421/F6.jpg", "media/2096f52618a3171db421/F7.jpg", "media/2096f52618a3171db421/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fed7f52831e0bcd8eb5690bd6b9d90fabd761945524b757e41c75e8a5483c227"}
{"qa_id": "youtube_vis__6797f20aa9__object_world_motion_direction__01", "video_id": "youtube_vis__6797f20aa9", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object A (person in white shirt and blue jeans riding a green bike) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "right", "left", "down", "backward"], "frame_paths": ["media/ee7c032ddc35f72055f9/F0.jpg", "media/ee7c032ddc35f72055f9/F1.jpg", "media/ee7c032ddc35f72055f9/F2.jpg", "media/ee7c032ddc35f72055f9/F3.jpg", "media/ee7c032ddc35f72055f9/F4.jpg", "media/ee7c032ddc35f72055f9/F5.jpg", "media/ee7c032ddc35f72055f9/F6.jpg", "media/ee7c032ddc35f72055f9/F7.jpg", "media/ee7c032ddc35f72055f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object A (person in white shirt and blue jeans riding a green bike) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. right\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "04be8ae2222e85b9086f6cc8a8a25c8de0ece2949a38179036a38e19e4d72589"}
{"qa_id": "youtube_vis__6797f20aa9__camera_object_distance_change__02", "video_id": "youtube_vis__6797f20aa9", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, did object A (person in white shirt and blue jeans riding a green bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/ee7c032ddc35f72055f9/F0.jpg", "media/ee7c032ddc35f72055f9/F1.jpg", "media/ee7c032ddc35f72055f9/F2.jpg", "media/ee7c032ddc35f72055f9/F3.jpg", "media/ee7c032ddc35f72055f9/F4.jpg", "media/ee7c032ddc35f72055f9/F5.jpg", "media/ee7c032ddc35f72055f9/F6.jpg", "media/ee7c032ddc35f72055f9/F7.jpg", "media/ee7c032ddc35f72055f9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object A (person in white shirt and blue jeans riding a green bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "26a50a3cc246239d79560546b857aa431f0e18fe74a93457fa2349213166a71c"}
{"qa_id": "youtube_vis__6797f20aa9__object_world_motion_direction__03", "video_id": "youtube_vis__6797f20aa9", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F6, in which dominant direction did object B (person in black shirt and green shorts riding a green bike) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "backward", "forward", "up"], "frame_paths": ["media/aedb68bf942c60bb5793/F0.jpg", "media/aedb68bf942c60bb5793/F1.jpg", "media/aedb68bf942c60bb5793/F2.jpg", "media/aedb68bf942c60bb5793/F3.jpg", "media/aedb68bf942c60bb5793/F4.jpg", "media/aedb68bf942c60bb5793/F5.jpg", "media/aedb68bf942c60bb5793/F6.jpg", "media/aedb68bf942c60bb5793/F7.jpg", "media/aedb68bf942c60bb5793/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F6, in which dominant direction did object B (person in black shirt and green shorts riding a green bike) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cca00d8647c6a2f4b23613a795668c3e559d4216eb66bbc243527950e239c114"}
{"qa_id": "youtube_vis__6797f20aa9__camera_object_distance_change__04", "video_id": "youtube_vis__6797f20aa9", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, did object B (person in black shirt and green shorts riding a green bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/aedb68bf942c60bb5793/F0.jpg", "media/aedb68bf942c60bb5793/F1.jpg", "media/aedb68bf942c60bb5793/F2.jpg", "media/aedb68bf942c60bb5793/F3.jpg", "media/aedb68bf942c60bb5793/F4.jpg", "media/aedb68bf942c60bb5793/F5.jpg", "media/aedb68bf942c60bb5793/F6.jpg", "media/aedb68bf942c60bb5793/F7.jpg", "media/aedb68bf942c60bb5793/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object B (person in black shirt and green shorts riding a green bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c691c4ed1937e4e336cea5cd3343fdb77b5aa61d5d86449cb5bdacffdaced7a"}
{"qa_id": "youtube_vis__684dc1c40c__camera_object_distance_change__00", "video_id": "youtube_vis__684dc1c40c", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object A (person wearing a red shirt and blue jeans riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/cc6e880e651e6413a271/F0.jpg", "media/cc6e880e651e6413a271/F1.jpg", "media/cc6e880e651e6413a271/F2.jpg", "media/cc6e880e651e6413a271/F3.jpg", "media/cc6e880e651e6413a271/F4.jpg", "media/cc6e880e651e6413a271/F5.jpg", "media/cc6e880e651e6413a271/F6.jpg", "media/cc6e880e651e6413a271/F7.jpg", "media/cc6e880e651e6413a271/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (person wearing a red shirt and blue jeans riding a horse) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bf71462bd23422f1a2c98f59e7061658f83f78a4a52f6c9a042de69eccc2d5ef"}
{"qa_id": "youtube_vis__684dc1c40c__camera_object_distance_change__01", "video_id": "youtube_vis__684dc1c40c", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (brown horse with a rider) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a68af0c8441c11a5bfac/F0.jpg", "media/a68af0c8441c11a5bfac/F1.jpg", "media/a68af0c8441c11a5bfac/F2.jpg", "media/a68af0c8441c11a5bfac/F3.jpg", "media/a68af0c8441c11a5bfac/F4.jpg", "media/a68af0c8441c11a5bfac/F5.jpg", "media/a68af0c8441c11a5bfac/F6.jpg", "media/a68af0c8441c11a5bfac/F7.jpg", "media/a68af0c8441c11a5bfac/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (brown horse with a rider) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b050b644d4110fd9b9b17c55bfa129739a83c1e9e0a468d1f0b2af9a65d1a70c"}
{"qa_id": "youtube_vis__68807d8601__object_world_motion_direction__00", "video_id": "youtube_vis__68807d8601", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F6, in which dominant direction did object A (small white speedboat creating a wake) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "left", "down", "up", "right", "forward"], "frame_paths": ["media/550e5f8d937608dd0aff/F0.jpg", "media/550e5f8d937608dd0aff/F1.jpg", "media/550e5f8d937608dd0aff/F2.jpg", "media/550e5f8d937608dd0aff/F3.jpg", "media/550e5f8d937608dd0aff/F4.jpg", "media/550e5f8d937608dd0aff/F5.jpg", "media/550e5f8d937608dd0aff/F6.jpg", "media/550e5f8d937608dd0aff/F7.jpg", "media/550e5f8d937608dd0aff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F6, in which dominant direction did object A (small white speedboat creating a wake) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. left\nC. down\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f1f00d8eb33ba187472c26e61dc4a5c967333a56a834d3bb061fc004bb89fddf"}
{"qa_id": "youtube_vis__68807d8601__camera_object_distance_change__01", "video_id": "youtube_vis__68807d8601", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F6, did object A (small white speedboat creating a wake) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/550e5f8d937608dd0aff/F0.jpg", "media/550e5f8d937608dd0aff/F1.jpg", "media/550e5f8d937608dd0aff/F2.jpg", "media/550e5f8d937608dd0aff/F3.jpg", "media/550e5f8d937608dd0aff/F4.jpg", "media/550e5f8d937608dd0aff/F5.jpg", "media/550e5f8d937608dd0aff/F6.jpg", "media/550e5f8d937608dd0aff/F7.jpg", "media/550e5f8d937608dd0aff/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, did object A (small white speedboat creating a wake) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8afa0b02d0ae850df523ec5651cea7cbf71861f604115bcd76190727f20a1d32"}
{"qa_id": "youtube_vis__695dcb6f16__object_world_motion_direction__00", "video_id": "youtube_vis__695dcb6f16", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object B (person wearing a high-visibility yellow vest and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "up", "right", "down", "backward", "forward"], "frame_paths": ["media/30d006cd87a6decadd6d/F0.jpg", "media/30d006cd87a6decadd6d/F1.jpg", "media/30d006cd87a6decadd6d/F2.jpg", "media/30d006cd87a6decadd6d/F3.jpg", "media/30d006cd87a6decadd6d/F4.jpg", "media/30d006cd87a6decadd6d/F5.jpg", "media/30d006cd87a6decadd6d/F6.jpg", "media/30d006cd87a6decadd6d/F7.jpg", "media/30d006cd87a6decadd6d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object B (person wearing a high-visibility yellow vest and dark pants) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. right\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "25c482f779a097192a5fbe0b1435c48c964dbeb6aa50b2ed182688ffa240e877"}
{"qa_id": "youtube_vis__6acc6049d9__camera_translation_direction__00", "video_id": "youtube_vis__6acc6049d9", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "backward", "down", "right", "left"], "frame_paths": ["media/4b9b34c3de0a55bfa793/F0.jpg", "media/4b9b34c3de0a55bfa793/F1.jpg", "media/4b9b34c3de0a55bfa793/F2.jpg", "media/4b9b34c3de0a55bfa793/F3.jpg", "media/4b9b34c3de0a55bfa793/F4.jpg", "media/4b9b34c3de0a55bfa793/F5.jpg", "media/4b9b34c3de0a55bfa793/F6.jpg", "media/4b9b34c3de0a55bfa793/F7.jpg", "media/4b9b34c3de0a55bfa793/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. backward\nD. down\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "87d1bd31224df4204f5571060e6096110b8721772f5e5362a65977d4c511c7a7"}
{"qa_id": "youtube_vis__6acc6049d9__camera_object_distance_change__01", "video_id": "youtube_vis__6acc6049d9", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (black pug with a pink collar) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/5a894cb2cd4c3d18cc9d/F0.jpg", "media/5a894cb2cd4c3d18cc9d/F1.jpg", "media/5a894cb2cd4c3d18cc9d/F2.jpg", "media/5a894cb2cd4c3d18cc9d/F3.jpg", "media/5a894cb2cd4c3d18cc9d/F4.jpg", "media/5a894cb2cd4c3d18cc9d/F5.jpg", "media/5a894cb2cd4c3d18cc9d/F6.jpg", "media/5a894cb2cd4c3d18cc9d/F7.jpg", "media/5a894cb2cd4c3d18cc9d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (black pug with a pink collar) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "92310c2ee0951f76613d687b763355f0e43a6ea460c78027b15f82d4a38a1a67"}
{"qa_id": "youtube_vis__6aea2205ec__object_travel_distance_comparison__00", "video_id": "youtube_vis__6aea2205ec", "task": "object_travel_distance_comparison", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (snowboarder wearing a grey jacket and black pants) or B (light green snowboard)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/45864bba2352e5f0140d/F0.jpg", "media/45864bba2352e5f0140d/F1.jpg", "media/45864bba2352e5f0140d/F2.jpg", "media/45864bba2352e5f0140d/F3.jpg", "media/45864bba2352e5f0140d/F4.jpg", "media/45864bba2352e5f0140d/F5.jpg", "media/45864bba2352e5f0140d/F6.jpg", "media/45864bba2352e5f0140d/F7.jpg", "media/45864bba2352e5f0140d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (snowboarder wearing a grey jacket and black pants) or B (light green snowboard)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a3eda2600b526aad8d0efcf2f85d3bb65901e97721896a2e001bdba0e0609f5d"}
{"qa_id": "youtube_vis__6af7af671f__object_world_motion_direction__00", "video_id": "youtube_vis__6af7af671f", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "down", "forward", "backward", "up", "right"], "frame_paths": ["media/c3aeff4141edfed98e67/F0.jpg", "media/c3aeff4141edfed98e67/F1.jpg", "media/c3aeff4141edfed98e67/F2.jpg", "media/c3aeff4141edfed98e67/F3.jpg", "media/c3aeff4141edfed98e67/F4.jpg", "media/c3aeff4141edfed98e67/F5.jpg", "media/c3aeff4141edfed98e67/F6.jpg", "media/c3aeff4141edfed98e67/F7.jpg", "media/c3aeff4141edfed98e67/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F4, in which dominant direction did object A (person in orange shirt) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. forward\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8d84a6662efc97f8445f66c7d411c8f959076c40a69b759bb0ea07693be1ac22"}
{"qa_id": "youtube_vis__6af7af671f__camera_object_distance_change__01", "video_id": "youtube_vis__6af7af671f", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F1 to F4, did object A (person in orange shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c3aeff4141edfed98e67/F0.jpg", "media/c3aeff4141edfed98e67/F1.jpg", "media/c3aeff4141edfed98e67/F2.jpg", "media/c3aeff4141edfed98e67/F3.jpg", "media/c3aeff4141edfed98e67/F4.jpg", "media/c3aeff4141edfed98e67/F5.jpg", "media/c3aeff4141edfed98e67/F6.jpg", "media/c3aeff4141edfed98e67/F7.jpg", "media/c3aeff4141edfed98e67/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (person in orange shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "440f1adc41847add705c106fe6df899e491c388750bdb4b25f7944b3e70185d4"}
{"qa_id": "youtube_vis__6bf95df2b9__object_world_motion_direction__00", "video_id": "youtube_vis__6bf95df2b9", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in grey shirt and dark pants skateboarding) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["up", "right", "forward", "backward", "down", "left"], "frame_paths": ["media/0c454135a0278122c536/F0.jpg", "media/0c454135a0278122c536/F1.jpg", "media/0c454135a0278122c536/F2.jpg", "media/0c454135a0278122c536/F3.jpg", "media/0c454135a0278122c536/F4.jpg", "media/0c454135a0278122c536/F5.jpg", "media/0c454135a0278122c536/F6.jpg", "media/0c454135a0278122c536/F7.jpg", "media/0c454135a0278122c536/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in grey shirt and dark pants skateboarding) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "059fc80a727b7f5efa8da62db23c1a67663f48c4d5daed3c21cb05bb9aa94bb9"}
{"qa_id": "youtube_vis__6bf95df2b9__camera_object_distance_change__01", "video_id": "youtube_vis__6bf95df2b9", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F7, did object A (person in grey shirt and dark pants skateboarding) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/0c454135a0278122c536/F0.jpg", "media/0c454135a0278122c536/F1.jpg", "media/0c454135a0278122c536/F2.jpg", "media/0c454135a0278122c536/F3.jpg", "media/0c454135a0278122c536/F4.jpg", "media/0c454135a0278122c536/F5.jpg", "media/0c454135a0278122c536/F6.jpg", "media/0c454135a0278122c536/F7.jpg", "media/0c454135a0278122c536/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, did object A (person in grey shirt and dark pants skateboarding) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f729a7a1354cea68e232e0b8282f5ce95eb3d35c05fc438fcd66e275e8b79e4"}
{"qa_id": "youtube_vis__6cc791250b__camera_translation_direction__00", "video_id": "youtube_vis__6cc791250b", "task": "camera_translation_direction", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "down", "left", "up", "right"], "frame_paths": ["media/2f6e7033121b08cff513/F0.jpg", "media/2f6e7033121b08cff513/F1.jpg", "media/2f6e7033121b08cff513/F2.jpg", "media/2f6e7033121b08cff513/F3.jpg", "media/2f6e7033121b08cff513/F4.jpg", "media/2f6e7033121b08cff513/F5.jpg", "media/2f6e7033121b08cff513/F6.jpg", "media/2f6e7033121b08cff513/F7.jpg", "media/2f6e7033121b08cff513/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. down\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c822e1bc7ccb4cceccdab8876e76736fbdf5b7690866957a5347a6a7b5c6f6e2"}
{"qa_id": "youtube_vis__6cc791250b__object_pair_distance_change__01", "video_id": "youtube_vis__6cc791250b", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did the 3D distance between object A (man in a dark suit) and object B (lion with a light-colored mane) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/a934d7869fcbc4c4199f/F0.jpg", "media/a934d7869fcbc4c4199f/F1.jpg", "media/a934d7869fcbc4c4199f/F2.jpg", "media/a934d7869fcbc4c4199f/F3.jpg", "media/a934d7869fcbc4c4199f/F4.jpg", "media/a934d7869fcbc4c4199f/F5.jpg", "media/a934d7869fcbc4c4199f/F6.jpg", "media/a934d7869fcbc4c4199f/F7.jpg", "media/a934d7869fcbc4c4199f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did the 3D distance between object A (man in a dark suit) and object B (lion with a light-colored mane) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "618cdd21e8ed20b02f7f687d6e5780f24acc40455531ab3cfd7fb12e3a037be0"}
{"qa_id": "youtube_vis__6cedf0aabd__object_pair_distance_change__00", "video_id": "youtube_vis__6cedf0aabd", "task": "object_pair_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F5 to F8, did the 3D distance between object A (person in black shirt and orange shorts) and object B (person in white shirt and green shorts) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/31f324d9f6c790ea4273/F0.jpg", "media/31f324d9f6c790ea4273/F1.jpg", "media/31f324d9f6c790ea4273/F2.jpg", "media/31f324d9f6c790ea4273/F3.jpg", "media/31f324d9f6c790ea4273/F4.jpg", "media/31f324d9f6c790ea4273/F5.jpg", "media/31f324d9f6c790ea4273/F6.jpg", "media/31f324d9f6c790ea4273/F7.jpg", "media/31f324d9f6c790ea4273/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g4"}, "requested_object_ids": ["g2", "g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did the 3D distance between object A (person in black shirt and orange shorts) and object B (person in white shirt and green shorts) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b3851076b9762053b3874e74cc0d03b19ad27e8a4dc413e93abca906ca664ee1"}
{"qa_id": "youtube_vis__6d04ed1fcd__object_world_motion_direction__00", "video_id": "youtube_vis__6d04ed1fcd", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (white BMW rally car with number 91 and Prince decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "up", "forward", "left", "down", "backward"], "frame_paths": ["media/51a455faf2a22f5e449c/F0.jpg", "media/51a455faf2a22f5e449c/F1.jpg", "media/51a455faf2a22f5e449c/F2.jpg", "media/51a455faf2a22f5e449c/F3.jpg", "media/51a455faf2a22f5e449c/F4.jpg", "media/51a455faf2a22f5e449c/F5.jpg", "media/51a455faf2a22f5e449c/F6.jpg", "media/51a455faf2a22f5e449c/F7.jpg", "media/51a455faf2a22f5e449c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (white BMW rally car with number 91 and Prince decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. forward\nD. left\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "826811cf16ab2ef198d97d4a17215e1b441ce92df0ee3d8163d97309133e7b5b"}
{"qa_id": "youtube_vis__6d6fb37d94__camera_translation_direction__00", "video_id": "youtube_vis__6d6fb37d94", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "down", "forward", "right", "left"], "frame_paths": ["media/04ecb9b5ab5deefdc867/F0.jpg", "media/04ecb9b5ab5deefdc867/F1.jpg", "media/04ecb9b5ab5deefdc867/F2.jpg", "media/04ecb9b5ab5deefdc867/F3.jpg", "media/04ecb9b5ab5deefdc867/F4.jpg", "media/04ecb9b5ab5deefdc867/F5.jpg", "media/04ecb9b5ab5deefdc867/F6.jpg", "media/04ecb9b5ab5deefdc867/F7.jpg", "media/04ecb9b5ab5deefdc867/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. down\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ceb490b6b8643f21727758cc97b8eaeade45588deb8c7fee12dbbdca660e4fb9"}
{"qa_id": "youtube_vis__6eb2e1cd9e__camera_view_direction_change__00", "video_id": "youtube_vis__6eb2e1cd9e", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/030a5ca78f7640f6dcd4/F0.jpg", "media/030a5ca78f7640f6dcd4/F1.jpg", "media/030a5ca78f7640f6dcd4/F2.jpg", "media/030a5ca78f7640f6dcd4/F3.jpg", "media/030a5ca78f7640f6dcd4/F4.jpg", "media/030a5ca78f7640f6dcd4/F5.jpg", "media/030a5ca78f7640f6dcd4/F6.jpg", "media/030a5ca78f7640f6dcd4/F7.jpg", "media/030a5ca78f7640f6dcd4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "53dde6c27cc780697bb33173c2cb51346c7596d4333e8108843d2086f5466f9a"}
{"qa_id": "youtube_vis__6eb2e1cd9e__camera_object_distance_change__01", "video_id": "youtube_vis__6eb2e1cd9e", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (snowboarder in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/de978df59f5fa08e6be5/F0.jpg", "media/de978df59f5fa08e6be5/F1.jpg", "media/de978df59f5fa08e6be5/F2.jpg", "media/de978df59f5fa08e6be5/F3.jpg", "media/de978df59f5fa08e6be5/F4.jpg", "media/de978df59f5fa08e6be5/F5.jpg", "media/de978df59f5fa08e6be5/F6.jpg", "media/de978df59f5fa08e6be5/F7.jpg", "media/de978df59f5fa08e6be5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (snowboarder in dark clothing) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dd15cd9d2adbbdd757e78645bf8fb5d4723d74f91d8c884ac05d8639ee75b9c4"}
{"qa_id": "youtube_vis__70ec0d96b9__camera_view_direction_change__00", "video_id": "youtube_vis__70ec0d96b9", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/b6c8c232a54a2fc3f682/F0.jpg", "media/b6c8c232a54a2fc3f682/F1.jpg", "media/b6c8c232a54a2fc3f682/F2.jpg", "media/b6c8c232a54a2fc3f682/F3.jpg", "media/b6c8c232a54a2fc3f682/F4.jpg", "media/b6c8c232a54a2fc3f682/F5.jpg", "media/b6c8c232a54a2fc3f682/F6.jpg", "media/b6c8c232a54a2fc3f682/F7.jpg", "media/b6c8c232a54a2fc3f682/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g6"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9927cdfa87aaa0ce88ba6cb5d19ce5e16240cff3a7c4a7c84d1379e5d3f4844d"}
{"qa_id": "youtube_vis__711dce6fe2__camera_translation_direction__00", "video_id": "youtube_vis__711dce6fe2", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "forward", "backward", "left"], "frame_paths": ["media/0f595af91b6ebb62da33/F0.jpg", "media/0f595af91b6ebb62da33/F1.jpg", "media/0f595af91b6ebb62da33/F2.jpg", "media/0f595af91b6ebb62da33/F3.jpg", "media/0f595af91b6ebb62da33/F4.jpg", "media/0f595af91b6ebb62da33/F5.jpg", "media/0f595af91b6ebb62da33/F6.jpg", "media/0f595af91b6ebb62da33/F7.jpg", "media/0f595af91b6ebb62da33/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8ff6c22b39396c53e2af88cdfbf1eedc99b7d3d61dfe0f8b870e1b707911c39c"}
{"qa_id": "youtube_vis__711dce6fe2__camera_view_direction_change__01", "video_id": "youtube_vis__711dce6fe2", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/0f595af91b6ebb62da33/F0.jpg", "media/0f595af91b6ebb62da33/F1.jpg", "media/0f595af91b6ebb62da33/F2.jpg", "media/0f595af91b6ebb62da33/F3.jpg", "media/0f595af91b6ebb62da33/F4.jpg", "media/0f595af91b6ebb62da33/F5.jpg", "media/0f595af91b6ebb62da33/F6.jpg", "media/0f595af91b6ebb62da33/F7.jpg", "media/0f595af91b6ebb62da33/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ce25bb87bb620ace6099607ac1c71ca17d20ef2f539ebd85ec0da7ac4a79421a"}
{"qa_id": "youtube_vis__711dce6fe2__camera_object_distance_change__02", "video_id": "youtube_vis__711dce6fe2", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F8, did object A (person in a white striped shirt seen from behind) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/6399e0142a6115a96775/F0.jpg", "media/6399e0142a6115a96775/F1.jpg", "media/6399e0142a6115a96775/F2.jpg", "media/6399e0142a6115a96775/F3.jpg", "media/6399e0142a6115a96775/F4.jpg", "media/6399e0142a6115a96775/F5.jpg", "media/6399e0142a6115a96775/F6.jpg", "media/6399e0142a6115a96775/F7.jpg", "media/6399e0142a6115a96775/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F8, did object A (person in a white striped shirt seen from behind) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78c3a8dbc57fd87c9deebb375183e1d475351570e69b14b64f84aeeb47502278"}
{"qa_id": "youtube_vis__72e8d1c1ff__camera_translation_direction__00", "video_id": "youtube_vis__72e8d1c1ff", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "forward", "left"], "frame_paths": ["media/e83f89a4c5901097413e/F0.jpg", "media/e83f89a4c5901097413e/F1.jpg", "media/e83f89a4c5901097413e/F2.jpg", "media/e83f89a4c5901097413e/F3.jpg", "media/e83f89a4c5901097413e/F4.jpg", "media/e83f89a4c5901097413e/F5.jpg", "media/e83f89a4c5901097413e/F6.jpg", "media/e83f89a4c5901097413e/F7.jpg", "media/e83f89a4c5901097413e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e8a2382ffacb3858d186bdee9218d4f446c23d22448870acb8dde5f8b14a4c7b"}
{"qa_id": "youtube_vis__72e8d1c1ff__camera_view_direction_change__01", "video_id": "youtube_vis__72e8d1c1ff", "task": "camera_view_direction_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/e83f89a4c5901097413e/F0.jpg", "media/e83f89a4c5901097413e/F1.jpg", "media/e83f89a4c5901097413e/F2.jpg", "media/e83f89a4c5901097413e/F3.jpg", "media/e83f89a4c5901097413e/F4.jpg", "media/e83f89a4c5901097413e/F5.jpg", "media/e83f89a4c5901097413e/F6.jpg", "media/e83f89a4c5901097413e/F7.jpg", "media/e83f89a4c5901097413e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "youtube_vis__735a547379__camera_view_direction_change__00", "video_id": "youtube_vis__735a547379", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["up", "left", "right", "down"], "frame_paths": ["media/3dd0cff5e3d2f01c8b19/F0.jpg", "media/3dd0cff5e3d2f01c8b19/F1.jpg", "media/3dd0cff5e3d2f01c8b19/F2.jpg", "media/3dd0cff5e3d2f01c8b19/F3.jpg", "media/3dd0cff5e3d2f01c8b19/F4.jpg", "media/3dd0cff5e3d2f01c8b19/F5.jpg", "media/3dd0cff5e3d2f01c8b19/F6.jpg", "media/3dd0cff5e3d2f01c8b19/F7.jpg", "media/3dd0cff5e3d2f01c8b19/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. up\nB. left\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4cbd30249df03db84f5ed7bffd7ab30cdf72c9ac8b07d584a7a624e09495c664"}
{"qa_id": "youtube_vis__74d05e4305__object_world_motion_direction__00", "video_id": "youtube_vis__74d05e4305", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "up", "forward", "right"], "frame_paths": ["media/63d5d32c2c80f691056a/F0.jpg", "media/63d5d32c2c80f691056a/F1.jpg", "media/63d5d32c2c80f691056a/F2.jpg", "media/63d5d32c2c80f691056a/F3.jpg", "media/63d5d32c2c80f691056a/F4.jpg", "media/63d5d32c2c80f691056a/F5.jpg", "media/63d5d32c2c80f691056a/F6.jpg", "media/63d5d32c2c80f691056a/F7.jpg", "media/63d5d32c2c80f691056a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (black sedan) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. up\nE. forward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64f3686ab77d7ba5672b7beb48c50c314618ef26c13b7b51c3a752140e83ecb5"}
{"qa_id": "youtube_vis__74d05e4305__camera_object_distance_change__01", "video_id": "youtube_vis__74d05e4305", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/63d5d32c2c80f691056a/F0.jpg", "media/63d5d32c2c80f691056a/F1.jpg", "media/63d5d32c2c80f691056a/F2.jpg", "media/63d5d32c2c80f691056a/F3.jpg", "media/63d5d32c2c80f691056a/F4.jpg", "media/63d5d32c2c80f691056a/F5.jpg", "media/63d5d32c2c80f691056a/F6.jpg", "media/63d5d32c2c80f691056a/F7.jpg", "media/63d5d32c2c80f691056a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (black sedan) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1b0228262819bd3bf0b7f755dfef65d51eb648b43644ebbdc17be2b49542d3c5"}
{"qa_id": "youtube_vis__76eb3a9bdd__object_world_motion_direction__00", "video_id": "youtube_vis__76eb3a9bdd", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (man wearing a black t-shirt with a white graphic and a green cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "up", "forward"], "frame_paths": ["media/51ddc500293dc81bf375/F0.jpg", "media/51ddc500293dc81bf375/F1.jpg", "media/51ddc500293dc81bf375/F2.jpg", "media/51ddc500293dc81bf375/F3.jpg", "media/51ddc500293dc81bf375/F4.jpg", "media/51ddc500293dc81bf375/F5.jpg", "media/51ddc500293dc81bf375/F6.jpg", "media/51ddc500293dc81bf375/F7.jpg", "media/51ddc500293dc81bf375/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (man wearing a black t-shirt with a white graphic and a green cap) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "44fcb12895ec98fd54f78e8646e0a15d36556f1a09e1920507609902d3809988"}
{"qa_id": "youtube_vis__76eb3a9bdd__object_pair_distance_change__01", "video_id": "youtube_vis__76eb3a9bdd", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did the 3D distance between object A (orange tiger with black stripes walking on grass) and object B (man wearing a black t-shirt with a white graphic and a green cap) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/96b34cce6e527ce3aee0/F0.jpg", "media/96b34cce6e527ce3aee0/F1.jpg", "media/96b34cce6e527ce3aee0/F2.jpg", "media/96b34cce6e527ce3aee0/F3.jpg", "media/96b34cce6e527ce3aee0/F4.jpg", "media/96b34cce6e527ce3aee0/F5.jpg", "media/96b34cce6e527ce3aee0/F6.jpg", "media/96b34cce6e527ce3aee0/F7.jpg", "media/96b34cce6e527ce3aee0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did the 3D distance between object A (orange tiger with black stripes walking on grass) and object B (man wearing a black t-shirt with a white graphic and a green cap) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1be946bb6379c42750f6d41a0cff768ec1f5601c760bcc7fb598387cc0cfc3d7"}
{"qa_id": "youtube_vis__7a17878a33__object_world_motion_direction__00", "video_id": "youtube_vis__7a17878a33", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F4, in which dominant direction did object A (silver hatchback car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "right", "forward", "left"], "frame_paths": ["media/a8fedd05cb5f165aa6d3/F0.jpg", "media/a8fedd05cb5f165aa6d3/F1.jpg", "media/a8fedd05cb5f165aa6d3/F2.jpg", "media/a8fedd05cb5f165aa6d3/F3.jpg", "media/a8fedd05cb5f165aa6d3/F4.jpg", "media/a8fedd05cb5f165aa6d3/F5.jpg", "media/a8fedd05cb5f165aa6d3/F6.jpg", "media/a8fedd05cb5f165aa6d3/F7.jpg", "media/a8fedd05cb5f165aa6d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F4, in which dominant direction did object A (silver hatchback car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. right\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "be48c2fe0df565d40d1aa0ff43b2d4683434c43faa94e5c189cdad419de7b57e"}
{"qa_id": "youtube_vis__7a17878a33__camera_object_distance_change__01", "video_id": "youtube_vis__7a17878a33", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a8fedd05cb5f165aa6d3/F0.jpg", "media/a8fedd05cb5f165aa6d3/F1.jpg", "media/a8fedd05cb5f165aa6d3/F2.jpg", "media/a8fedd05cb5f165aa6d3/F3.jpg", "media/a8fedd05cb5f165aa6d3/F4.jpg", "media/a8fedd05cb5f165aa6d3/F5.jpg", "media/a8fedd05cb5f165aa6d3/F6.jpg", "media/a8fedd05cb5f165aa6d3/F7.jpg", "media/a8fedd05cb5f165aa6d3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (silver hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8ddc122894b2c19348d4cfd35474c60ac0fbaa8205383f95ca6c3e802e00192c"}
{"qa_id": "youtube_vis__7b17cc8dc9__camera_view_direction_change__00", "video_id": "youtube_vis__7b17cc8dc9", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/288b1acf04baeb6d8507/F0.jpg", "media/288b1acf04baeb6d8507/F1.jpg", "media/288b1acf04baeb6d8507/F2.jpg", "media/288b1acf04baeb6d8507/F3.jpg", "media/288b1acf04baeb6d8507/F4.jpg", "media/288b1acf04baeb6d8507/F5.jpg", "media/288b1acf04baeb6d8507/F6.jpg", "media/288b1acf04baeb6d8507/F7.jpg", "media/288b1acf04baeb6d8507/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b05c59c5ff5c9031ea68746d67c85c07217775e2ecc9e157cbe29c5a6487d912"}
{"qa_id": "youtube_vis__7e5d433323__camera_translation_direction__00", "video_id": "youtube_vis__7e5d433323", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "up", "forward", "backward"], "frame_paths": ["media/24fe5448372e8d0e4ad6/F0.jpg", "media/24fe5448372e8d0e4ad6/F1.jpg", "media/24fe5448372e8d0e4ad6/F2.jpg", "media/24fe5448372e8d0e4ad6/F3.jpg", "media/24fe5448372e8d0e4ad6/F4.jpg", "media/24fe5448372e8d0e4ad6/F5.jpg", "media/24fe5448372e8d0e4ad6/F6.jpg", "media/24fe5448372e8d0e4ad6/F7.jpg", "media/24fe5448372e8d0e4ad6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. up\nE. forward\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8c4dd2635133eb480441639762bab9f72eed2ed9da8bd64a3244fd7baab79c17"}
{"qa_id": "youtube_vis__7eb2605d96__camera_view_direction_change__00", "video_id": "youtube_vis__7eb2605d96", "task": "camera_view_direction_change", "group": "LLL", "group4": "LLL", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["down", "right", "up", "left"], "frame_paths": ["media/81e2a333ef5ef1c4cf0d/F0.jpg", "media/81e2a333ef5ef1c4cf0d/F1.jpg", "media/81e2a333ef5ef1c4cf0d/F2.jpg", "media/81e2a333ef5ef1c4cf0d/F3.jpg", "media/81e2a333ef5ef1c4cf0d/F4.jpg", "media/81e2a333ef5ef1c4cf0d/F5.jpg", "media/81e2a333ef5ef1c4cf0d/F6.jpg", "media/81e2a333ef5ef1c4cf0d/F7.jpg", "media/81e2a333ef5ef1c4cf0d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. down\nB. right\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1f9558cfcb39b4b8062b0685a5042b148145270b4c4b58790fc901bc14fc2ec8"}
{"qa_id": "youtube_vis__80b5e08983__camera_object_distance_change__00", "video_id": "youtube_vis__80b5e08983", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did object B (child in pink shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/dd57caa14a7bf7b169b0/F0.jpg", "media/dd57caa14a7bf7b169b0/F1.jpg", "media/dd57caa14a7bf7b169b0/F2.jpg", "media/dd57caa14a7bf7b169b0/F3.jpg", "media/dd57caa14a7bf7b169b0/F4.jpg", "media/dd57caa14a7bf7b169b0/F5.jpg", "media/dd57caa14a7bf7b169b0/F6.jpg", "media/dd57caa14a7bf7b169b0/F7.jpg", "media/dd57caa14a7bf7b169b0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object B (child in pink shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "710a7239edd4135d79947e2faa6c92d60ba3cb827d98f0ec0770f002a3c5e665"}
{"qa_id": "youtube_vis__80b5e08983__object_pair_distance_change__01", "video_id": "youtube_vis__80b5e08983", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, did the 3D distance between object A (brown horse grazing) and object B (child in pink shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/5c9121c1164c47926906/F0.jpg", "media/5c9121c1164c47926906/F1.jpg", "media/5c9121c1164c47926906/F2.jpg", "media/5c9121c1164c47926906/F3.jpg", "media/5c9121c1164c47926906/F4.jpg", "media/5c9121c1164c47926906/F5.jpg", "media/5c9121c1164c47926906/F6.jpg", "media/5c9121c1164c47926906/F7.jpg", "media/5c9121c1164c47926906/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did the 3D distance between object A (brown horse grazing) and object B (child in pink shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "df4a8e106dabad74f774da9ea9823c82fe0298f473f41eb85582a03cd7557cfa"}
{"qa_id": "youtube_vis__80b5e08983__object_travel_distance_comparison__02", "video_id": "youtube_vis__80b5e08983", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse grazing) or B (child in pink shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/5c9121c1164c47926906/F0.jpg", "media/5c9121c1164c47926906/F1.jpg", "media/5c9121c1164c47926906/F2.jpg", "media/5c9121c1164c47926906/F3.jpg", "media/5c9121c1164c47926906/F4.jpg", "media/5c9121c1164c47926906/F5.jpg", "media/5c9121c1164c47926906/F6.jpg", "media/5c9121c1164c47926906/F7.jpg", "media/5c9121c1164c47926906/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F8, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (brown horse grazing) or B (child in pink shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f108527c9c158aeabd8e031e98ce8f50801eb18c9832e36d0ddb073e39846708"}
{"qa_id": "youtube_vis__8157b1bc1f__object_travel_distance_comparison__00", "video_id": "youtube_vis__8157b1bc1f", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in brown shirt and grey shorts) or B (person in black shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/438a849b5d7dc9ee3908/F0.jpg", "media/438a849b5d7dc9ee3908/F1.jpg", "media/438a849b5d7dc9ee3908/F2.jpg", "media/438a849b5d7dc9ee3908/F3.jpg", "media/438a849b5d7dc9ee3908/F4.jpg", "media/438a849b5d7dc9ee3908/F5.jpg", "media/438a849b5d7dc9ee3908/F6.jpg", "media/438a849b5d7dc9ee3908/F7.jpg", "media/438a849b5d7dc9ee3908/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F2, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in brown shirt and grey shorts) or B (person in black shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5d7de19d1db40c927f64bfe82fa133e922b015526615b30c88287f59bf8968a0"}
{"qa_id": "youtube_vis__8200245704__camera_view_direction_change__00", "video_id": "youtube_vis__8200245704", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "down", "left", "right"], "frame_paths": ["media/d4455fb5ae62c721ed6b/F0.jpg", "media/d4455fb5ae62c721ed6b/F1.jpg", "media/d4455fb5ae62c721ed6b/F2.jpg", "media/d4455fb5ae62c721ed6b/F3.jpg", "media/d4455fb5ae62c721ed6b/F4.jpg", "media/d4455fb5ae62c721ed6b/F5.jpg", "media/d4455fb5ae62c721ed6b/F6.jpg", "media/d4455fb5ae62c721ed6b/F7.jpg", "media/d4455fb5ae62c721ed6b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. down\nC. left\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f112bd9e2150b78d414584be3560dcb538cb2ca4637fa74da960d29415cb12b"}
{"qa_id": "youtube_vis__8200245704__camera_object_distance_change__01", "video_id": "youtube_vis__8200245704", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object A (motorcyclist in white and black racing suit) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/81b0dbc2fcedeebdba09/F0.jpg", "media/81b0dbc2fcedeebdba09/F1.jpg", "media/81b0dbc2fcedeebdba09/F2.jpg", "media/81b0dbc2fcedeebdba09/F3.jpg", "media/81b0dbc2fcedeebdba09/F4.jpg", "media/81b0dbc2fcedeebdba09/F5.jpg", "media/81b0dbc2fcedeebdba09/F6.jpg", "media/81b0dbc2fcedeebdba09/F7.jpg", "media/81b0dbc2fcedeebdba09/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object A (motorcyclist in white and black racing suit) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a489d03374fa14860942c954b418ecba679dd402c8f51a33183490050c057315"}
{"qa_id": "youtube_vis__8200245704__camera_object_distance_change__02", "video_id": "youtube_vis__8200245704", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F7, did object B (green dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/099314e208b803eea1fc/F0.jpg", "media/099314e208b803eea1fc/F1.jpg", "media/099314e208b803eea1fc/F2.jpg", "media/099314e208b803eea1fc/F3.jpg", "media/099314e208b803eea1fc/F4.jpg", "media/099314e208b803eea1fc/F5.jpg", "media/099314e208b803eea1fc/F6.jpg", "media/099314e208b803eea1fc/F7.jpg", "media/099314e208b803eea1fc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F7, did object B (green dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1e5f4550e19bb5e87c73693719638c6f4c460d18ad7b1a9f745e2aa69653dbc7"}
{"qa_id": "youtube_vis__837ea3be6a__camera_object_distance_change__00", "video_id": "youtube_vis__837ea3be6a", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (shirtless person riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/6a9ed2761dc96810befd/F0.jpg", "media/6a9ed2761dc96810befd/F1.jpg", "media/6a9ed2761dc96810befd/F2.jpg", "media/6a9ed2761dc96810befd/F3.jpg", "media/6a9ed2761dc96810befd/F4.jpg", "media/6a9ed2761dc96810befd/F5.jpg", "media/6a9ed2761dc96810befd/F6.jpg", "media/6a9ed2761dc96810befd/F7.jpg", "media/6a9ed2761dc96810befd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (shirtless person riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "051b55742136f670795e67a3322e4b1c9f3055ffe4ce21306f5dd047e1682047"}
{"qa_id": "youtube_vis__837ea3be6a__camera_object_distance_change__01", "video_id": "youtube_vis__837ea3be6a", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (bicycle with green frame) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/60953af690d1d6f255a4/F0.jpg", "media/60953af690d1d6f255a4/F1.jpg", "media/60953af690d1d6f255a4/F2.jpg", "media/60953af690d1d6f255a4/F3.jpg", "media/60953af690d1d6f255a4/F4.jpg", "media/60953af690d1d6f255a4/F5.jpg", "media/60953af690d1d6f255a4/F6.jpg", "media/60953af690d1d6f255a4/F7.jpg", "media/60953af690d1d6f255a4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (bicycle with green frame) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7fd240d43d62a6c7756d6c943978f329b1fe0766075217ac0883b7fe7e4e910f"}
{"qa_id": "youtube_vis__84a4bf147d__object_world_motion_direction__00", "video_id": "youtube_vis__84a4bf147d", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in yellow shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/94864cd033f6516298a5/F0.jpg", "media/94864cd033f6516298a5/F1.jpg", "media/94864cd033f6516298a5/F2.jpg", "media/94864cd033f6516298a5/F3.jpg", "media/94864cd033f6516298a5/F4.jpg", "media/94864cd033f6516298a5/F5.jpg", "media/94864cd033f6516298a5/F6.jpg", "media/94864cd033f6516298a5/F7.jpg", "media/94864cd033f6516298a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in yellow shirt) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1c4510c9c961c7dd24aa731dd70d28c860ef0ffe99a22bab48e5c0ef2a84953b"}
{"qa_id": "youtube_vis__8554d8b755__object_world_motion_direction__00", "video_id": "youtube_vis__8554d8b755", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in black shirt and shorts riding a bicycle) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "up", "right", "left", "down"], "frame_paths": ["media/8bb94708f14509dc253e/F0.jpg", "media/8bb94708f14509dc253e/F1.jpg", "media/8bb94708f14509dc253e/F2.jpg", "media/8bb94708f14509dc253e/F3.jpg", "media/8bb94708f14509dc253e/F4.jpg", "media/8bb94708f14509dc253e/F5.jpg", "media/8bb94708f14509dc253e/F6.jpg", "media/8bb94708f14509dc253e/F7.jpg", "media/8bb94708f14509dc253e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (person in black shirt and shorts riding a bicycle) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. up\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a647de405746a8454567d134c1ee6a2b87b4ee240a44bb0c4ca1d1288001bb0a"}
{"qa_id": "youtube_vis__8554d8b755__object_world_motion_direction__01", "video_id": "youtube_vis__8554d8b755", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object B (black bicycle with white wheels) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["right", "down", "forward", "left", "backward", "up"], "frame_paths": ["media/8f29d86edc509103fb76/F0.jpg", "media/8f29d86edc509103fb76/F1.jpg", "media/8f29d86edc509103fb76/F2.jpg", "media/8f29d86edc509103fb76/F3.jpg", "media/8f29d86edc509103fb76/F4.jpg", "media/8f29d86edc509103fb76/F5.jpg", "media/8f29d86edc509103fb76/F6.jpg", "media/8f29d86edc509103fb76/F7.jpg", "media/8f29d86edc509103fb76/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object B (black bicycle with white wheels) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. forward\nD. left\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bb661961dca7110deecba2b584bc54f1d5c4d7b4c07a896bf86bd4808b2c7aa1"}
{"qa_id": "youtube_vis__8a39cc3d11__object_world_motion_direction__00", "video_id": "youtube_vis__8a39cc3d11", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (giraffe with orange contour walking left) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "forward", "left", "up", "right"], "frame_paths": ["media/01b5244b0f1233005e46/F0.jpg", "media/01b5244b0f1233005e46/F1.jpg", "media/01b5244b0f1233005e46/F2.jpg", "media/01b5244b0f1233005e46/F3.jpg", "media/01b5244b0f1233005e46/F4.jpg", "media/01b5244b0f1233005e46/F5.jpg", "media/01b5244b0f1233005e46/F6.jpg", "media/01b5244b0f1233005e46/F7.jpg", "media/01b5244b0f1233005e46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (giraffe with orange contour walking left) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. forward\nD. left\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "087de9f314fc146a4537fb241246ad813e6ca2224332b808bdeb6d7f1598cc94"}
{"qa_id": "youtube_vis__8a39cc3d11__object_world_motion_direction__01", "video_id": "youtube_vis__8a39cc3d11", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object B (giraffe with green contour walking right) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/fcf075b44a4a933886b9/F0.jpg", "media/fcf075b44a4a933886b9/F1.jpg", "media/fcf075b44a4a933886b9/F2.jpg", "media/fcf075b44a4a933886b9/F3.jpg", "media/fcf075b44a4a933886b9/F4.jpg", "media/fcf075b44a4a933886b9/F5.jpg", "media/fcf075b44a4a933886b9/F6.jpg", "media/fcf075b44a4a933886b9/F7.jpg", "media/fcf075b44a4a933886b9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object B (giraffe with green contour walking right) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "43978e6bdf85f3bd9bcd4333ce12d2d5550f5e97759ad11fb035250c393533d9"}
{"qa_id": "youtube_vis__8aa733a4c2__camera_view_direction_change__00", "video_id": "youtube_vis__8aa733a4c2", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "right", "up"], "frame_paths": ["media/cba6585b962a80a973a9/F0.jpg", "media/cba6585b962a80a973a9/F1.jpg", "media/cba6585b962a80a973a9/F2.jpg", "media/cba6585b962a80a973a9/F3.jpg", "media/cba6585b962a80a973a9/F4.jpg", "media/cba6585b962a80a973a9/F5.jpg", "media/cba6585b962a80a973a9/F6.jpg", "media/cba6585b962a80a973a9/F7.jpg", "media/cba6585b962a80a973a9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. right\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d2ebf55124b9ea1c872fd57169bbfc6bc3facb58f344120f30f6db226bab927d"}
{"qa_id": "youtube_vis__8aca214360__camera_view_direction_change__00", "video_id": "youtube_vis__8aca214360", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/b5ff366ef76d944d4ae6/F0.jpg", "media/b5ff366ef76d944d4ae6/F1.jpg", "media/b5ff366ef76d944d4ae6/F2.jpg", "media/b5ff366ef76d944d4ae6/F3.jpg", "media/b5ff366ef76d944d4ae6/F4.jpg", "media/b5ff366ef76d944d4ae6/F5.jpg", "media/b5ff366ef76d944d4ae6/F6.jpg", "media/b5ff366ef76d944d4ae6/F7.jpg", "media/b5ff366ef76d944d4ae6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "64fa65743ff5458c37df2b8880c0ab0a341602cee3eec7a95aaa58c83a978348"}
{"qa_id": "youtube_vis__8d44170b57__camera_object_distance_change__00", "video_id": "youtube_vis__8d44170b57", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object A (white dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9dab1a67d1e4a6c4c159/F0.jpg", "media/9dab1a67d1e4a6c4c159/F1.jpg", "media/9dab1a67d1e4a6c4c159/F2.jpg", "media/9dab1a67d1e4a6c4c159/F3.jpg", "media/9dab1a67d1e4a6c4c159/F4.jpg", "media/9dab1a67d1e4a6c4c159/F5.jpg", "media/9dab1a67d1e4a6c4c159/F6.jpg", "media/9dab1a67d1e4a6c4c159/F7.jpg", "media/9dab1a67d1e4a6c4c159/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (white dirt bike) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8126bcb785795ce1e9a8a6e8f65ed0cda6a73e0a4df19dea2bff8d6287a8a52b"}
{"qa_id": "youtube_vis__8d44170b57__object_world_motion_direction__01", "video_id": "youtube_vis__8d44170b57", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (rider in green shirt and white helmet) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "up", "backward", "right", "forward", "down"], "frame_paths": ["media/87d04d077ba30dde128f/F0.jpg", "media/87d04d077ba30dde128f/F1.jpg", "media/87d04d077ba30dde128f/F2.jpg", "media/87d04d077ba30dde128f/F3.jpg", "media/87d04d077ba30dde128f/F4.jpg", "media/87d04d077ba30dde128f/F5.jpg", "media/87d04d077ba30dde128f/F6.jpg", "media/87d04d077ba30dde128f/F7.jpg", "media/87d04d077ba30dde128f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (rider in green shirt and white helmet) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. up\nC. backward\nD. right\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5dbc98529164c6ebf5cfa99a1a7e862c342cfa579b3d960db803db4fa91f20f3"}
{"qa_id": "youtube_vis__8d44170b57__camera_object_distance_change__02", "video_id": "youtube_vis__8d44170b57", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object B (rider in green shirt and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/87d04d077ba30dde128f/F0.jpg", "media/87d04d077ba30dde128f/F1.jpg", "media/87d04d077ba30dde128f/F2.jpg", "media/87d04d077ba30dde128f/F3.jpg", "media/87d04d077ba30dde128f/F4.jpg", "media/87d04d077ba30dde128f/F5.jpg", "media/87d04d077ba30dde128f/F6.jpg", "media/87d04d077ba30dde128f/F7.jpg", "media/87d04d077ba30dde128f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object B (rider in green shirt and white helmet) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d1970f211a049701be8ddbd5a0c54294742e23f1fc398968b9449d5d068ae344"}
{"qa_id": "youtube_vis__8e5cbc711b__camera_translation_direction__00", "video_id": "youtube_vis__8e5cbc711b", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "up", "right", "backward", "left"], "frame_paths": ["media/ed4e88f1ea76f75cced5/F0.jpg", "media/ed4e88f1ea76f75cced5/F1.jpg", "media/ed4e88f1ea76f75cced5/F2.jpg", "media/ed4e88f1ea76f75cced5/F3.jpg", "media/ed4e88f1ea76f75cced5/F4.jpg", "media/ed4e88f1ea76f75cced5/F5.jpg", "media/ed4e88f1ea76f75cced5/F6.jpg", "media/ed4e88f1ea76f75cced5/F7.jpg", "media/ed4e88f1ea76f75cced5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. up\nD. right\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e574bf57e861d2462b526968c1ce6dbceb93a735c3ac117a1b493951314e67d9"}
{"qa_id": "youtube_vis__8e5cbc711b__object_world_motion_direction__01", "video_id": "youtube_vis__8e5cbc711b", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (dark grey flatbed truck driving on a snowy dirt road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "forward", "left", "down", "up"], "frame_paths": ["media/a49da44618c51df3aff7/F0.jpg", "media/a49da44618c51df3aff7/F1.jpg", "media/a49da44618c51df3aff7/F2.jpg", "media/a49da44618c51df3aff7/F3.jpg", "media/a49da44618c51df3aff7/F4.jpg", "media/a49da44618c51df3aff7/F5.jpg", "media/a49da44618c51df3aff7/F6.jpg", "media/a49da44618c51df3aff7/F7.jpg", "media/a49da44618c51df3aff7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (dark grey flatbed truck driving on a snowy dirt road) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. forward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "761c80f5db15b0d5c644a1c18f21f74e3be830d9226971067828cd554335c171"}
{"qa_id": "youtube_vis__8e5cbc711b__camera_object_distance_change__02", "video_id": "youtube_vis__8e5cbc711b", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F8, did object A (dark grey flatbed truck driving on a snowy dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a49da44618c51df3aff7/F0.jpg", "media/a49da44618c51df3aff7/F1.jpg", "media/a49da44618c51df3aff7/F2.jpg", "media/a49da44618c51df3aff7/F3.jpg", "media/a49da44618c51df3aff7/F4.jpg", "media/a49da44618c51df3aff7/F5.jpg", "media/a49da44618c51df3aff7/F6.jpg", "media/a49da44618c51df3aff7/F7.jpg", "media/a49da44618c51df3aff7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, did object A (dark grey flatbed truck driving on a snowy dirt road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "11e013548990501681bb83cad45d0a7439ce32ea5e04463372ffb30d99473557"}
{"qa_id": "youtube_vis__90273e63e1__object_world_motion_direction__00", "video_id": "youtube_vis__90273e63e1", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (green rally car with white and red decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "right", "down", "left"], "frame_paths": ["media/b14f26c182732ed7164b/F0.jpg", "media/b14f26c182732ed7164b/F1.jpg", "media/b14f26c182732ed7164b/F2.jpg", "media/b14f26c182732ed7164b/F3.jpg", "media/b14f26c182732ed7164b/F4.jpg", "media/b14f26c182732ed7164b/F5.jpg", "media/b14f26c182732ed7164b/F6.jpg", "media/b14f26c182732ed7164b/F7.jpg", "media/b14f26c182732ed7164b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (green rally car with white and red decals) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6cc78e73161608125c7a4df11fa0edf38122beb73a5980da19a7985264670b2c"}
{"qa_id": "youtube_vis__90273e63e1__camera_object_distance_change__01", "video_id": "youtube_vis__90273e63e1", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F8, did object A (green rally car with white and red decals) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/b14f26c182732ed7164b/F0.jpg", "media/b14f26c182732ed7164b/F1.jpg", "media/b14f26c182732ed7164b/F2.jpg", "media/b14f26c182732ed7164b/F3.jpg", "media/b14f26c182732ed7164b/F4.jpg", "media/b14f26c182732ed7164b/F5.jpg", "media/b14f26c182732ed7164b/F6.jpg", "media/b14f26c182732ed7164b/F7.jpg", "media/b14f26c182732ed7164b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, did object A (green rally car with white and red decals) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65347dfb1af744eafced90c15b0333b225aa7b2546e1c1492b3644373fa9cab8"}
{"qa_id": "youtube_vis__911a08836d__camera_translation_direction__00", "video_id": "youtube_vis__911a08836d", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "up", "right", "down", "left"], "frame_paths": ["media/8da6d090b05cb796412f/F0.jpg", "media/8da6d090b05cb796412f/F1.jpg", "media/8da6d090b05cb796412f/F2.jpg", "media/8da6d090b05cb796412f/F3.jpg", "media/8da6d090b05cb796412f/F4.jpg", "media/8da6d090b05cb796412f/F5.jpg", "media/8da6d090b05cb796412f/F6.jpg", "media/8da6d090b05cb796412f/F7.jpg", "media/8da6d090b05cb796412f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. up\nD. right\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b8621bfd1aa305da1fed2d1b452bc62f2ad056a459767dcc4634ad8b1d7c120d"}
{"qa_id": "youtube_vis__946d71fb5d__camera_view_direction_change__00", "video_id": "youtube_vis__946d71fb5d", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/82512fe18eba3ca6ad25/F0.jpg", "media/82512fe18eba3ca6ad25/F1.jpg", "media/82512fe18eba3ca6ad25/F2.jpg", "media/82512fe18eba3ca6ad25/F3.jpg", "media/82512fe18eba3ca6ad25/F4.jpg", "media/82512fe18eba3ca6ad25/F5.jpg", "media/82512fe18eba3ca6ad25/F6.jpg", "media/82512fe18eba3ca6ad25/F7.jpg", "media/82512fe18eba3ca6ad25/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F4?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f33a29b4c20a7ca1a9178075cc426b67db59f1387b0efa4d6bf164475fd35c63"}
{"qa_id": "youtube_vis__946d71fb5d__camera_object_distance_change__01", "video_id": "youtube_vis__946d71fb5d", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F3, did object B (dark grey horse with a rider in a red shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/8d296ba7802a6155b935/F0.jpg", "media/8d296ba7802a6155b935/F1.jpg", "media/8d296ba7802a6155b935/F2.jpg", "media/8d296ba7802a6155b935/F3.jpg", "media/8d296ba7802a6155b935/F4.jpg", "media/8d296ba7802a6155b935/F5.jpg", "media/8d296ba7802a6155b935/F6.jpg", "media/8d296ba7802a6155b935/F7.jpg", "media/8d296ba7802a6155b935/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object B (dark grey horse with a rider in a red shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "afb55fbfc72e7e4f8894418b82a4fb32e4450f9fed27298eab119d2e25916685"}
{"qa_id": "youtube_vis__946d71fb5d__object_travel_distance_comparison__02", "video_id": "youtube_vis__946d71fb5d", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in grey shirt and orange pants holding a whip) or B (dark grey horse with a rider in a red shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["A", "B"], "frame_paths": ["media/e111685a3b9250acd383/F0.jpg", "media/e111685a3b9250acd383/F1.jpg", "media/e111685a3b9250acd383/F2.jpg", "media/e111685a3b9250acd383/F3.jpg", "media/e111685a3b9250acd383/F4.jpg", "media/e111685a3b9250acd383/F5.jpg", "media/e111685a3b9250acd383/F6.jpg", "media/e111685a3b9250acd383/F7.jpg", "media/e111685a3b9250acd383/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F3, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (person in grey shirt and orange pants holding a whip) or B (dark grey horse with a rider in a red shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object A\nB. Object B\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "bf8a05d7d5236fa9eaa548e287d6b83a96546b2797d81b6e4481094077f8c763"}
{"qa_id": "youtube_vis__997117a654__object_world_motion_direction__00", "video_id": "youtube_vis__997117a654", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in white t-shirt and dark pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "up", "right", "forward"], "frame_paths": ["media/750043c7728745db6c6a/F0.jpg", "media/750043c7728745db6c6a/F1.jpg", "media/750043c7728745db6c6a/F2.jpg", "media/750043c7728745db6c6a/F3.jpg", "media/750043c7728745db6c6a/F4.jpg", "media/750043c7728745db6c6a/F5.jpg", "media/750043c7728745db6c6a/F6.jpg", "media/750043c7728745db6c6a/F7.jpg", "media/750043c7728745db6c6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (person in white t-shirt and dark pants) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4515b9e9bb086dec6689d2a8210b531a95c06b0782245d5dceec4fc7777d9924"}
{"qa_id": "youtube_vis__997117a654__camera_object_distance_change__01", "video_id": "youtube_vis__997117a654", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, did object A (person in white t-shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/750043c7728745db6c6a/F0.jpg", "media/750043c7728745db6c6a/F1.jpg", "media/750043c7728745db6c6a/F2.jpg", "media/750043c7728745db6c6a/F3.jpg", "media/750043c7728745db6c6a/F4.jpg", "media/750043c7728745db6c6a/F5.jpg", "media/750043c7728745db6c6a/F6.jpg", "media/750043c7728745db6c6a/F7.jpg", "media/750043c7728745db6c6a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (person in white t-shirt and dark pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d137dc8f59a0c2c3ceba048df6837413fa28434b0f9d3b232098618ce3b90818"}
{"qa_id": "youtube_vis__99c4277aee__object_world_motion_direction__00", "video_id": "youtube_vis__99c4277aee", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F7, in which dominant direction did object A (black sedan race car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["down", "up", "right", "left", "backward", "forward"], "frame_paths": ["media/3858ef0c5b10f15df949/F0.jpg", "media/3858ef0c5b10f15df949/F1.jpg", "media/3858ef0c5b10f15df949/F2.jpg", "media/3858ef0c5b10f15df949/F3.jpg", "media/3858ef0c5b10f15df949/F4.jpg", "media/3858ef0c5b10f15df949/F5.jpg", "media/3858ef0c5b10f15df949/F6.jpg", "media/3858ef0c5b10f15df949/F7.jpg", "media/3858ef0c5b10f15df949/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F7, in which dominant direction did object A (black sedan race car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. right\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4f340ef37ae835bcd176ceb962137af5c9ab611b75430df32e8461593c6d7041"}
{"qa_id": "youtube_vis__99c4277aee__camera_object_distance_change__01", "video_id": "youtube_vis__99c4277aee", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, did object A (black sedan race car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/3858ef0c5b10f15df949/F0.jpg", "media/3858ef0c5b10f15df949/F1.jpg", "media/3858ef0c5b10f15df949/F2.jpg", "media/3858ef0c5b10f15df949/F3.jpg", "media/3858ef0c5b10f15df949/F4.jpg", "media/3858ef0c5b10f15df949/F5.jpg", "media/3858ef0c5b10f15df949/F6.jpg", "media/3858ef0c5b10f15df949/F7.jpg", "media/3858ef0c5b10f15df949/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, did object A (black sedan race car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1986a0471780f4ee9aac5ad84ce832441a67ed5bdf34a96d2347bbf0620c84b8"}
{"qa_id": "youtube_vis__99c6b1acf2__camera_object_distance_change__00", "video_id": "youtube_vis__99c6b1acf2", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did object A (small motorboat with a person in a yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/93e89ce21d5baa0bbee9/F0.jpg", "media/93e89ce21d5baa0bbee9/F1.jpg", "media/93e89ce21d5baa0bbee9/F2.jpg", "media/93e89ce21d5baa0bbee9/F3.jpg", "media/93e89ce21d5baa0bbee9/F4.jpg", "media/93e89ce21d5baa0bbee9/F5.jpg", "media/93e89ce21d5baa0bbee9/F6.jpg", "media/93e89ce21d5baa0bbee9/F7.jpg", "media/93e89ce21d5baa0bbee9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (small motorboat with a person in a yellow shirt) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b83a4decbf1c7c06f34bce322f9cc83ac13381723268d5d9e01b819d30ec2d5f"}
{"qa_id": "youtube_vis__9a17abd0f3__camera_object_distance_change__00", "video_id": "youtube_vis__9a17abd0f3", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F4 to F8, did object A (cyclist wearing a blue and white jersey) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/541fc70e035aa10d492b/F0.jpg", "media/541fc70e035aa10d492b/F1.jpg", "media/541fc70e035aa10d492b/F2.jpg", "media/541fc70e035aa10d492b/F3.jpg", "media/541fc70e035aa10d492b/F4.jpg", "media/541fc70e035aa10d492b/F5.jpg", "media/541fc70e035aa10d492b/F6.jpg", "media/541fc70e035aa10d492b/F7.jpg", "media/541fc70e035aa10d492b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (cyclist wearing a blue and white jersey) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3c89a5dcfbf33f4addbe327b9c7dbf942e3ab1986521a237bac324d3bcd50fa3"}
{"qa_id": "youtube_vis__9a878c9390__object_world_motion_direction__00", "video_id": "youtube_vis__9a878c9390", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (shirtless man in black shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "backward", "left", "down", "forward"], "frame_paths": ["media/fefc3473f4496a37cd6e/F0.jpg", "media/fefc3473f4496a37cd6e/F1.jpg", "media/fefc3473f4496a37cd6e/F2.jpg", "media/fefc3473f4496a37cd6e/F3.jpg", "media/fefc3473f4496a37cd6e/F4.jpg", "media/fefc3473f4496a37cd6e/F5.jpg", "media/fefc3473f4496a37cd6e/F6.jpg", "media/fefc3473f4496a37cd6e/F7.jpg", "media/fefc3473f4496a37cd6e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (shirtless man in black shorts) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. backward\nD. left\nE. down\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1c29575d7064db3765ecd1bf7b5d083160261ca24bd42e75e78f12a3c3ac96af"}
{"qa_id": "youtube_vis__9a878c9390__camera_object_distance_change__01", "video_id": "youtube_vis__9a878c9390", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F2, did object A (shirtless man in black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/fefc3473f4496a37cd6e/F0.jpg", "media/fefc3473f4496a37cd6e/F1.jpg", "media/fefc3473f4496a37cd6e/F2.jpg", "media/fefc3473f4496a37cd6e/F3.jpg", "media/fefc3473f4496a37cd6e/F4.jpg", "media/fefc3473f4496a37cd6e/F5.jpg", "media/fefc3473f4496a37cd6e/F6.jpg", "media/fefc3473f4496a37cd6e/F7.jpg", "media/fefc3473f4496a37cd6e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (shirtless man in black shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5bc6f7c4099b03fd75d9521ff9bf391fd50128b118bfc15a20082011fd9a7e06"}
{"qa_id": "youtube_vis__a0ad437be4__camera_translation_direction__00", "video_id": "youtube_vis__a0ad437be4", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "forward", "right", "up"], "frame_paths": ["media/f79428ad1b5a2e997713/F0.jpg", "media/f79428ad1b5a2e997713/F1.jpg", "media/f79428ad1b5a2e997713/F2.jpg", "media/f79428ad1b5a2e997713/F3.jpg", "media/f79428ad1b5a2e997713/F4.jpg", "media/f79428ad1b5a2e997713/F5.jpg", "media/f79428ad1b5a2e997713/F6.jpg", "media/f79428ad1b5a2e997713/F7.jpg", "media/f79428ad1b5a2e997713/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. forward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "814884b0a3e679464dc35ee59550d7a75498a627c2d0a2890d3076edc12c7165"}
{"qa_id": "youtube_vis__a0ad437be4__camera_object_distance_change__01", "video_id": "youtube_vis__a0ad437be4", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, did object A (person in white shirt and shorts riding a wooden board down a grassy hill) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/6be8e8fe8c6c6961b6a5/F0.jpg", "media/6be8e8fe8c6c6961b6a5/F1.jpg", "media/6be8e8fe8c6c6961b6a5/F2.jpg", "media/6be8e8fe8c6c6961b6a5/F3.jpg", "media/6be8e8fe8c6c6961b6a5/F4.jpg", "media/6be8e8fe8c6c6961b6a5/F5.jpg", "media/6be8e8fe8c6c6961b6a5/F6.jpg", "media/6be8e8fe8c6c6961b6a5/F7.jpg", "media/6be8e8fe8c6c6961b6a5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (person in white shirt and shorts riding a wooden board down a grassy hill) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d827eb4808a9db70141c5470ea3923a95eb234b8e8f9805966719191323f4c7c"}
{"qa_id": "youtube_vis__a2c996e429__camera_translation_direction__00", "video_id": "youtube_vis__a2c996e429", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "up", "down", "left", "forward"], "frame_paths": ["media/092bdb836fd73a0f1a0a/F0.jpg", "media/092bdb836fd73a0f1a0a/F1.jpg", "media/092bdb836fd73a0f1a0a/F2.jpg", "media/092bdb836fd73a0f1a0a/F3.jpg", "media/092bdb836fd73a0f1a0a/F4.jpg", "media/092bdb836fd73a0f1a0a/F5.jpg", "media/092bdb836fd73a0f1a0a/F6.jpg", "media/092bdb836fd73a0f1a0a/F7.jpg", "media/092bdb836fd73a0f1a0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. up\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c2bf8911f377960da0d5b0f5cb409d9201652d9081674d18d5d166abbabdf1c8"}
{"qa_id": "youtube_vis__a2c996e429__object_world_motion_direction__01", "video_id": "youtube_vis__a2c996e429", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object A (white bus with a large advertisement featuring a smiling woman on the rear) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "backward", "forward", "up"], "frame_paths": ["media/bcc69f9231e7860b1d46/F0.jpg", "media/bcc69f9231e7860b1d46/F1.jpg", "media/bcc69f9231e7860b1d46/F2.jpg", "media/bcc69f9231e7860b1d46/F3.jpg", "media/bcc69f9231e7860b1d46/F4.jpg", "media/bcc69f9231e7860b1d46/F5.jpg", "media/bcc69f9231e7860b1d46/F6.jpg", "media/bcc69f9231e7860b1d46/F7.jpg", "media/bcc69f9231e7860b1d46/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object A (white bus with a large advertisement featuring a smiling woman on the rear) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ea538faaeef29f813867d5f95713ab3f4d939d706a939b43b9bf0dc5ca65b790"}
{"qa_id": "youtube_vis__a2c996e429__object_world_motion_direction__02", "video_id": "youtube_vis__a2c996e429", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F6, in which dominant direction did object B (silver sedan driving on the right side of the road) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/e29a1c6d374723d54840/F0.jpg", "media/e29a1c6d374723d54840/F1.jpg", "media/e29a1c6d374723d54840/F2.jpg", "media/e29a1c6d374723d54840/F3.jpg", "media/e29a1c6d374723d54840/F4.jpg", "media/e29a1c6d374723d54840/F5.jpg", "media/e29a1c6d374723d54840/F6.jpg", "media/e29a1c6d374723d54840/F7.jpg", "media/e29a1c6d374723d54840/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F6, in which dominant direction did object B (silver sedan driving on the right side of the road) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6bebb28d970277baa555bf6b51e99da6d6a7df91e17b84d990c296ea2ba26445"}
{"qa_id": "youtube_vis__a2c996e429__camera_object_distance_change__03", "video_id": "youtube_vis__a2c996e429", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did object B (silver sedan driving on the right side of the road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/e29a1c6d374723d54840/F0.jpg", "media/e29a1c6d374723d54840/F1.jpg", "media/e29a1c6d374723d54840/F2.jpg", "media/e29a1c6d374723d54840/F3.jpg", "media/e29a1c6d374723d54840/F4.jpg", "media/e29a1c6d374723d54840/F5.jpg", "media/e29a1c6d374723d54840/F6.jpg", "media/e29a1c6d374723d54840/F7.jpg", "media/e29a1c6d374723d54840/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object B (silver sedan driving on the right side of the road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8d502e3a4892d82765992a40279c25ab72ef3de4fc0ea8189cf87d8cfbc66271"}
{"qa_id": "youtube_vis__a2c996e429__object_pair_distance_change__04", "video_id": "youtube_vis__a2c996e429", "task": "object_pair_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F6, did the 3D distance between object A (white bus with a large advertisement featuring a smiling woman on the rear) and object B (silver sedan driving on the right side of the road) increase or decrease? Compare their separation at the two specified frames.", "options": ["increased", "decreased"], "frame_paths": ["media/75d89c0f79da8a6c4fbe/F0.jpg", "media/75d89c0f79da8a6c4fbe/F1.jpg", "media/75d89c0f79da8a6c4fbe/F2.jpg", "media/75d89c0f79da8a6c4fbe/F3.jpg", "media/75d89c0f79da8a6c4fbe/F4.jpg", "media/75d89c0f79da8a6c4fbe/F5.jpg", "media/75d89c0f79da8a6c4fbe/F6.jpg", "media/75d89c0f79da8a6c4fbe/F7.jpg", "media/75d89c0f79da8a6c4fbe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did the 3D distance between object A (white bus with a large advertisement featuring a smiling woman on the rear) and object B (silver sedan driving on the right side of the road) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. increased\nB. decreased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee87251cd00d488f769037563f0d6fb3813e0f6260379b836f320be8ff82a0a2"}
{"qa_id": "youtube_vis__a2c996e429__object_travel_distance_comparison__05", "video_id": "youtube_vis__a2c996e429", "task": "object_travel_distance_comparison", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white bus with a large advertisement featuring a smiling woman on the rear) or B (silver sedan driving on the right side of the road)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/75d89c0f79da8a6c4fbe/F0.jpg", "media/75d89c0f79da8a6c4fbe/F1.jpg", "media/75d89c0f79da8a6c4fbe/F2.jpg", "media/75d89c0f79da8a6c4fbe/F3.jpg", "media/75d89c0f79da8a6c4fbe/F4.jpg", "media/75d89c0f79da8a6c4fbe/F5.jpg", "media/75d89c0f79da8a6c4fbe/F6.jpg", "media/75d89c0f79da8a6c4fbe/F7.jpg", "media/75d89c0f79da8a6c4fbe/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F3 through F6, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white bus with a large advertisement featuring a smiling woman on the rear) or B (silver sedan driving on the right side of the road)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8355edec0ed8ea3474d706b3187c865b3c512118d2ceca68fe24714a0ef00139"}
{"qa_id": "youtube_vis__a4baeabc97__object_world_motion_direction__00", "video_id": "youtube_vis__a4baeabc97", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (China Airlines passenger jet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "left", "backward", "down"], "frame_paths": ["media/293753c1ff056d2e0083/F0.jpg", "media/293753c1ff056d2e0083/F1.jpg", "media/293753c1ff056d2e0083/F2.jpg", "media/293753c1ff056d2e0083/F3.jpg", "media/293753c1ff056d2e0083/F4.jpg", "media/293753c1ff056d2e0083/F5.jpg", "media/293753c1ff056d2e0083/F6.jpg", "media/293753c1ff056d2e0083/F7.jpg", "media/293753c1ff056d2e0083/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (China Airlines passenger jet) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. left\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9696af6140b7d3bec322a7fccabe56222842c4088fa3590a040de699b2db6fe7"}
{"qa_id": "youtube_vis__a50eb5a0ea__camera_view_direction_change__00", "video_id": "youtube_vis__a50eb5a0ea", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "up", "right", "down"], "frame_paths": ["media/5cf884c46e8f9f13e29f/F0.jpg", "media/5cf884c46e8f9f13e29f/F1.jpg", "media/5cf884c46e8f9f13e29f/F2.jpg", "media/5cf884c46e8f9f13e29f/F3.jpg", "media/5cf884c46e8f9f13e29f/F4.jpg", "media/5cf884c46e8f9f13e29f/F5.jpg", "media/5cf884c46e8f9f13e29f/F6.jpg", "media/5cf884c46e8f9f13e29f/F7.jpg", "media/5cf884c46e8f9f13e29f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F7, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. up\nC. right\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4677dcb1f828d098f52a5e63b3d70e6694b217bdeddfa921a10e29f3f6066d98"}
{"qa_id": "youtube_vis__a56d08782d__camera_translation_direction__00", "video_id": "youtube_vis__a56d08782d", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "backward", "forward", "right", "left"], "frame_paths": ["media/07d6075b62eb079c2f2e/F0.jpg", "media/07d6075b62eb079c2f2e/F1.jpg", "media/07d6075b62eb079c2f2e/F2.jpg", "media/07d6075b62eb079c2f2e/F3.jpg", "media/07d6075b62eb079c2f2e/F4.jpg", "media/07d6075b62eb079c2f2e/F5.jpg", "media/07d6075b62eb079c2f2e/F6.jpg", "media/07d6075b62eb079c2f2e/F7.jpg", "media/07d6075b62eb079c2f2e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. backward\nD. forward\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a56b7797f384ad7dade2b330669276bbafac861fd60380d2b6ecff3de7ac2629"}
{"qa_id": "youtube_vis__a9d6aa2506__object_world_motion_direction__00", "video_id": "youtube_vis__a9d6aa2506", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F5 to F7, in which dominant direction did object A (red muscle car) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "up", "down", "right"], "frame_paths": ["media/db1629b81228896b8643/F0.jpg", "media/db1629b81228896b8643/F1.jpg", "media/db1629b81228896b8643/F2.jpg", "media/db1629b81228896b8643/F3.jpg", "media/db1629b81228896b8643/F4.jpg", "media/db1629b81228896b8643/F5.jpg", "media/db1629b81228896b8643/F6.jpg", "media/db1629b81228896b8643/F7.jpg", "media/db1629b81228896b8643/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F5 to F7, in which dominant direction did object A (red muscle car) move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. up\nE. down\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2246e1d5dbb2b2a56db6463903bb3fc46f87ebfea5dc6a874fe4152ff7ae06a8"}
{"qa_id": "youtube_vis__a9d6aa2506__object_travel_distance_comparison__01", "video_id": "youtube_vis__a9d6aa2506", "task": "object_travel_distance_comparison", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "Over the sampled trajectory from F5 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (red muscle car) or B (person in dark t-shirt and jeans)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/fc3e79cf9b96064c88dc/F0.jpg", "media/fc3e79cf9b96064c88dc/F1.jpg", "media/fc3e79cf9b96064c88dc/F2.jpg", "media/fc3e79cf9b96064c88dc/F3.jpg", "media/fc3e79cf9b96064c88dc/F4.jpg", "media/fc3e79cf9b96064c88dc/F5.jpg", "media/fc3e79cf9b96064c88dc/F6.jpg", "media/fc3e79cf9b96064c88dc/F7.jpg", "media/fc3e79cf9b96064c88dc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g1", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F5 through F7, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (red muscle car) or B (person in dark t-shirt and jeans)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cf26ec9ffdd7db15138fd83ee570a04e3d0b297e9fd2a0708e0f43f9624b1c88"}
{"qa_id": "youtube_vis__abb50c8697__object_world_motion_direction__00", "video_id": "youtube_vis__abb50c8697", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object B (mallard duck with green head and white neck ring) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "forward", "up", "down", "left"], "frame_paths": ["media/7f97bf7e094176edb1da/F0.jpg", "media/7f97bf7e094176edb1da/F1.jpg", "media/7f97bf7e094176edb1da/F2.jpg", "media/7f97bf7e094176edb1da/F3.jpg", "media/7f97bf7e094176edb1da/F4.jpg", "media/7f97bf7e094176edb1da/F5.jpg", "media/7f97bf7e094176edb1da/F6.jpg", "media/7f97bf7e094176edb1da/F7.jpg", "media/7f97bf7e094176edb1da/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object B (mallard duck with green head and white neck ring) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. forward\nD. up\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "5958fc0a17c925693c5131a02b1c18909ef4d328390a7fabc752fb8540a9fc30"}
{"qa_id": "youtube_vis__abe61a11bb__camera_translation_direction__00", "video_id": "youtube_vis__abe61a11bb", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "right", "backward", "up", "down"], "frame_paths": ["media/1b78a3293ef71fe1eb18/F0.jpg", "media/1b78a3293ef71fe1eb18/F1.jpg", "media/1b78a3293ef71fe1eb18/F2.jpg", "media/1b78a3293ef71fe1eb18/F3.jpg", "media/1b78a3293ef71fe1eb18/F4.jpg", "media/1b78a3293ef71fe1eb18/F5.jpg", "media/1b78a3293ef71fe1eb18/F6.jpg", "media/1b78a3293ef71fe1eb18/F7.jpg", "media/1b78a3293ef71fe1eb18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. right\nD. backward\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d71433064a473cb0d643c49e2675452759a7db891f737d384dd0a1898aa46ea8"}
{"qa_id": "youtube_vis__abe61a11bb__camera_view_direction_change__01", "video_id": "youtube_vis__abe61a11bb", "task": "camera_view_direction_change", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["down", "right", "left", "up"], "frame_paths": ["media/1b78a3293ef71fe1eb18/F0.jpg", "media/1b78a3293ef71fe1eb18/F1.jpg", "media/1b78a3293ef71fe1eb18/F2.jpg", "media/1b78a3293ef71fe1eb18/F3.jpg", "media/1b78a3293ef71fe1eb18/F4.jpg", "media/1b78a3293ef71fe1eb18/F5.jpg", "media/1b78a3293ef71fe1eb18/F6.jpg", "media/1b78a3293ef71fe1eb18/F7.jpg", "media/1b78a3293ef71fe1eb18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. down\nB. right\nC. left\nD. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1ba651badb03ebcf75ad7ea8dcb29386770b74ec4a286d714ce4b229922af015"}
{"qa_id": "youtube_vis__acf2c4b745__camera_view_direction_change__00", "video_id": "youtube_vis__acf2c4b745", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "down", "up", "left"], "frame_paths": ["media/9329f8d391f4a13d0306/F0.jpg", "media/9329f8d391f4a13d0306/F1.jpg", "media/9329f8d391f4a13d0306/F2.jpg", "media/9329f8d391f4a13d0306/F3.jpg", "media/9329f8d391f4a13d0306/F4.jpg", "media/9329f8d391f4a13d0306/F5.jpg", "media/9329f8d391f4a13d0306/F6.jpg", "media/9329f8d391f4a13d0306/F7.jpg", "media/9329f8d391f4a13d0306/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. down\nC. up\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "82fb06dddece22d028c76a1de500fe7f0467ea6c1091f07daafc7c36a5581ad7"}
{"qa_id": "youtube_vis__acf2c4b745__camera_object_distance_change__01", "video_id": "youtube_vis__acf2c4b745", "task": "camera_object_distance_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F2 to F4, did object A (person in white shirt and orange shorts riding a skateboard) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a28fc53540e535e4c8aa/F0.jpg", "media/a28fc53540e535e4c8aa/F1.jpg", "media/a28fc53540e535e4c8aa/F2.jpg", "media/a28fc53540e535e4c8aa/F3.jpg", "media/a28fc53540e535e4c8aa/F4.jpg", "media/a28fc53540e535e4c8aa/F5.jpg", "media/a28fc53540e535e4c8aa/F6.jpg", "media/a28fc53540e535e4c8aa/F7.jpg", "media/a28fc53540e535e4c8aa/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, did object A (person in white shirt and orange shorts riding a skateboard) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "00298d5e865c2b1d2fd5dd2a624dae4a3dd24051b248d7c3a408d7949483add2"}
{"qa_id": "youtube_vis__b53f675641__camera_translation_direction__00", "video_id": "youtube_vis__b53f675641", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "forward", "left", "down", "right", "up"], "frame_paths": ["media/2e254a99cf7f72cb5094/F0.jpg", "media/2e254a99cf7f72cb5094/F1.jpg", "media/2e254a99cf7f72cb5094/F2.jpg", "media/2e254a99cf7f72cb5094/F3.jpg", "media/2e254a99cf7f72cb5094/F4.jpg", "media/2e254a99cf7f72cb5094/F5.jpg", "media/2e254a99cf7f72cb5094/F6.jpg", "media/2e254a99cf7f72cb5094/F7.jpg", "media/2e254a99cf7f72cb5094/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. forward\nC. left\nD. down\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "73f47e644d46363d92c657c9fac0dd38615b87cb5fb44e0758dfc2ae2b635e08"}
{"qa_id": "youtube_vis__b53f675641__camera_object_distance_change__01", "video_id": "youtube_vis__b53f675641", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, did object A (person in a white shirt and dark vest parachuting) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/352eb374376e49725e01/F0.jpg", "media/352eb374376e49725e01/F1.jpg", "media/352eb374376e49725e01/F2.jpg", "media/352eb374376e49725e01/F3.jpg", "media/352eb374376e49725e01/F4.jpg", "media/352eb374376e49725e01/F5.jpg", "media/352eb374376e49725e01/F6.jpg", "media/352eb374376e49725e01/F7.jpg", "media/352eb374376e49725e01/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person in a white shirt and dark vest parachuting) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1c2e97b135256d66acd44167aacd2d5f073a41fe552f3d58f5ab171ba8535a7a"}
{"qa_id": "youtube_vis__b58df6732c__object_world_motion_direction__00", "video_id": "youtube_vis__b58df6732c", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (brown and white dog) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "up", "left", "forward", "right", "backward"], "frame_paths": ["media/38b3dd1402b304ad3268/F0.jpg", "media/38b3dd1402b304ad3268/F1.jpg", "media/38b3dd1402b304ad3268/F2.jpg", "media/38b3dd1402b304ad3268/F3.jpg", "media/38b3dd1402b304ad3268/F4.jpg", "media/38b3dd1402b304ad3268/F5.jpg", "media/38b3dd1402b304ad3268/F6.jpg", "media/38b3dd1402b304ad3268/F7.jpg", "media/38b3dd1402b304ad3268/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (brown and white dog) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. up\nC. left\nD. forward\nE. right\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f3d29df07c1f454b487e3af79b77f679b2d1e6027fb889801bff759482a8ba9a"}
{"qa_id": "youtube_vis__b58df6732c__camera_object_distance_change__01", "video_id": "youtube_vis__b58df6732c", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (brown and white dog) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/38b3dd1402b304ad3268/F0.jpg", "media/38b3dd1402b304ad3268/F1.jpg", "media/38b3dd1402b304ad3268/F2.jpg", "media/38b3dd1402b304ad3268/F3.jpg", "media/38b3dd1402b304ad3268/F4.jpg", "media/38b3dd1402b304ad3268/F5.jpg", "media/38b3dd1402b304ad3268/F6.jpg", "media/38b3dd1402b304ad3268/F7.jpg", "media/38b3dd1402b304ad3268/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (brown and white dog) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b663142a58ec4679176bc82d0555bff95fefd4c623930b9b0d703d9eb4d68b06"}
{"qa_id": "youtube_vis__b84b8ae665__camera_translation_direction__00", "video_id": "youtube_vis__b84b8ae665", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "left", "down", "up", "forward"], "frame_paths": ["media/9a9a4990cb2680ea1561/F0.jpg", "media/9a9a4990cb2680ea1561/F1.jpg", "media/9a9a4990cb2680ea1561/F2.jpg", "media/9a9a4990cb2680ea1561/F3.jpg", "media/9a9a4990cb2680ea1561/F4.jpg", "media/9a9a4990cb2680ea1561/F5.jpg", "media/9a9a4990cb2680ea1561/F6.jpg", "media/9a9a4990cb2680ea1561/F7.jpg", "media/9a9a4990cb2680ea1561/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. left\nD. down\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "930c1a64abea53620175115f0af315e80b5f9347debae80cffa6934acb3ec792"}
{"qa_id": "youtube_vis__bb1b0ee89f__camera_object_distance_change__00", "video_id": "youtube_vis__bb1b0ee89f", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F6, did object B (shirtless man sitting in the back of a boat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/6ced13b9f8ad448dccfb/F0.jpg", "media/6ced13b9f8ad448dccfb/F1.jpg", "media/6ced13b9f8ad448dccfb/F2.jpg", "media/6ced13b9f8ad448dccfb/F3.jpg", "media/6ced13b9f8ad448dccfb/F4.jpg", "media/6ced13b9f8ad448dccfb/F5.jpg", "media/6ced13b9f8ad448dccfb/F6.jpg", "media/6ced13b9f8ad448dccfb/F7.jpg", "media/6ced13b9f8ad448dccfb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F6, did object B (shirtless man sitting in the back of a boat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7921f0990d4d1d99e946b744f7b14a67aa8baaeccb5457839f49e8e6bd4175ae"}
{"qa_id": "youtube_vis__bcd1d39afb__camera_translation_direction__00", "video_id": "youtube_vis__bcd1d39afb", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "forward", "up", "left", "backward", "down"], "frame_paths": ["media/41d5c9d3dd1a5efce053/F0.jpg", "media/41d5c9d3dd1a5efce053/F1.jpg", "media/41d5c9d3dd1a5efce053/F2.jpg", "media/41d5c9d3dd1a5efce053/F3.jpg", "media/41d5c9d3dd1a5efce053/F4.jpg", "media/41d5c9d3dd1a5efce053/F5.jpg", "media/41d5c9d3dd1a5efce053/F6.jpg", "media/41d5c9d3dd1a5efce053/F7.jpg", "media/41d5c9d3dd1a5efce053/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. forward\nC. up\nD. left\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fa471174f2eae7278ac9cfb9ec30dc824bcd069c25c79c42b87174bde60e6997"}
{"qa_id": "youtube_vis__bcd1d39afb__camera_object_distance_change__01", "video_id": "youtube_vis__bcd1d39afb", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, did object A (silver bus with route number 265B displayed on the front) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/6c9d3b1635fbc66e9e63/F0.jpg", "media/6c9d3b1635fbc66e9e63/F1.jpg", "media/6c9d3b1635fbc66e9e63/F2.jpg", "media/6c9d3b1635fbc66e9e63/F3.jpg", "media/6c9d3b1635fbc66e9e63/F4.jpg", "media/6c9d3b1635fbc66e9e63/F5.jpg", "media/6c9d3b1635fbc66e9e63/F6.jpg", "media/6c9d3b1635fbc66e9e63/F7.jpg", "media/6c9d3b1635fbc66e9e63/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (silver bus with route number 265B displayed on the front) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d91c0fbc5b939a296783af24d85971e0dd64486cd32b29ac5509b94e975f3b5b"}
{"qa_id": "youtube_vis__c51821b990__object_world_motion_direction__00", "video_id": "youtube_vis__c51821b990", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in olive green shirt and dark shorts holding a pink object) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "right", "up", "backward", "down"], "frame_paths": ["media/394a1707d1ea8c3bfde3/F0.jpg", "media/394a1707d1ea8c3bfde3/F1.jpg", "media/394a1707d1ea8c3bfde3/F2.jpg", "media/394a1707d1ea8c3bfde3/F3.jpg", "media/394a1707d1ea8c3bfde3/F4.jpg", "media/394a1707d1ea8c3bfde3/F5.jpg", "media/394a1707d1ea8c3bfde3/F6.jpg", "media/394a1707d1ea8c3bfde3/F7.jpg", "media/394a1707d1ea8c3bfde3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object A (person in olive green shirt and dark shorts holding a pink object) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. right\nD. up\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3e14d355f1cec52e8f675eae66b0fabe92115e3e23dfd82f90d4d0ebeae8cccc"}
{"qa_id": "youtube_vis__c51821b990__camera_object_distance_change__01", "video_id": "youtube_vis__c51821b990", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F1 to F3, did object A (person in olive green shirt and dark shorts holding a pink object) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/394a1707d1ea8c3bfde3/F0.jpg", "media/394a1707d1ea8c3bfde3/F1.jpg", "media/394a1707d1ea8c3bfde3/F2.jpg", "media/394a1707d1ea8c3bfde3/F3.jpg", "media/394a1707d1ea8c3bfde3/F4.jpg", "media/394a1707d1ea8c3bfde3/F5.jpg", "media/394a1707d1ea8c3bfde3/F6.jpg", "media/394a1707d1ea8c3bfde3/F7.jpg", "media/394a1707d1ea8c3bfde3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F3, did object A (person in olive green shirt and dark shorts holding a pink object) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "78a536090bc48ecde102696dff44ed21b31a7c292cf58671ffb398b1abb4c176"}
{"qa_id": "youtube_vis__c6c8368e32__camera_translation_direction__00", "video_id": "youtube_vis__c6c8368e32", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "backward", "up", "right", "down"], "frame_paths": ["media/6d19fa06ca0c60b0a8ce/F0.jpg", "media/6d19fa06ca0c60b0a8ce/F1.jpg", "media/6d19fa06ca0c60b0a8ce/F2.jpg", "media/6d19fa06ca0c60b0a8ce/F3.jpg", "media/6d19fa06ca0c60b0a8ce/F4.jpg", "media/6d19fa06ca0c60b0a8ce/F5.jpg", "media/6d19fa06ca0c60b0a8ce/F6.jpg", "media/6d19fa06ca0c60b0a8ce/F7.jpg", "media/6d19fa06ca0c60b0a8ce/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. backward\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4dc2386f0a14f9583b4fda9e110663a25dc3252ba85660b709b005c7ccded329"}
{"qa_id": "youtube_vis__c6c8368e32__object_world_motion_direction__01", "video_id": "youtube_vis__c6c8368e32", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange car driving on the track) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["down", "backward", "right", "up", "forward", "left"], "frame_paths": ["media/613fb1c88a48fc44dc20/F0.jpg", "media/613fb1c88a48fc44dc20/F1.jpg", "media/613fb1c88a48fc44dc20/F2.jpg", "media/613fb1c88a48fc44dc20/F3.jpg", "media/613fb1c88a48fc44dc20/F4.jpg", "media/613fb1c88a48fc44dc20/F5.jpg", "media/613fb1c88a48fc44dc20/F6.jpg", "media/613fb1c88a48fc44dc20/F7.jpg", "media/613fb1c88a48fc44dc20/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (orange car driving on the track) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. backward\nC. right\nD. up\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "84f66bc65e3ff47394d25472dc6176c6171831c14e5ea9bb6adc262a5ffc25ea"}
{"qa_id": "youtube_vis__c7c2860db3__object_world_motion_direction__00", "video_id": "youtube_vis__c7c2860db3", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in white shirt and black shorts playing tennis) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "right", "up", "down", "backward"], "frame_paths": ["media/55b74e412caddc7f3ce7/F0.jpg", "media/55b74e412caddc7f3ce7/F1.jpg", "media/55b74e412caddc7f3ce7/F2.jpg", "media/55b74e412caddc7f3ce7/F3.jpg", "media/55b74e412caddc7f3ce7/F4.jpg", "media/55b74e412caddc7f3ce7/F5.jpg", "media/55b74e412caddc7f3ce7/F6.jpg", "media/55b74e412caddc7f3ce7/F7.jpg", "media/55b74e412caddc7f3ce7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F5, in which dominant direction did object A (person in white shirt and black shorts playing tennis) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. right\nD. up\nE. down\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e2443a99c34957d5009466c5f511e08dc2ab3f424b65f123104944e96213cc59"}
{"qa_id": "youtube_vis__ca4b99cbac__object_world_motion_direction__00", "video_id": "youtube_vis__ca4b99cbac", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F2 to F8, in which dominant direction did object A (zebra with black and white stripes) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/ac7b8c6dc47063c83b21/F0.jpg", "media/ac7b8c6dc47063c83b21/F1.jpg", "media/ac7b8c6dc47063c83b21/F2.jpg", "media/ac7b8c6dc47063c83b21/F3.jpg", "media/ac7b8c6dc47063c83b21/F4.jpg", "media/ac7b8c6dc47063c83b21/F5.jpg", "media/ac7b8c6dc47063c83b21/F6.jpg", "media/ac7b8c6dc47063c83b21/F7.jpg", "media/ac7b8c6dc47063c83b21/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F8, in which dominant direction did object A (zebra with black and white stripes) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f34b18638109cdfaadf24acd8de888bfbd432856d6f6b207eb3d233cd84d0a97"}
{"qa_id": "youtube_vis__cbb410da64__camera_translation_direction__00", "video_id": "youtube_vis__cbb410da64", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["up", "right", "forward", "backward", "down", "left"], "frame_paths": ["media/d28ddbb179b6f3fe5c5e/F0.jpg", "media/d28ddbb179b6f3fe5c5e/F1.jpg", "media/d28ddbb179b6f3fe5c5e/F2.jpg", "media/d28ddbb179b6f3fe5c5e/F3.jpg", "media/d28ddbb179b6f3fe5c5e/F4.jpg", "media/d28ddbb179b6f3fe5c5e/F5.jpg", "media/d28ddbb179b6f3fe5c5e/F6.jpg", "media/d28ddbb179b6f3fe5c5e/F7.jpg", "media/d28ddbb179b6f3fe5c5e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "07c89312a3f537ed1449fa24b2857b7dbfc9f0f95d213ec015ff643a64824525"}
{"qa_id": "youtube_vis__cfddad4954__object_world_motion_direction__00", "video_id": "youtube_vis__cfddad4954", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (silhouette of a person standing on the beach) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "forward", "backward", "up", "right", "left"], "frame_paths": ["media/86de67b5a9b212e615d7/F0.jpg", "media/86de67b5a9b212e615d7/F1.jpg", "media/86de67b5a9b212e615d7/F2.jpg", "media/86de67b5a9b212e615d7/F3.jpg", "media/86de67b5a9b212e615d7/F4.jpg", "media/86de67b5a9b212e615d7/F5.jpg", "media/86de67b5a9b212e615d7/F6.jpg", "media/86de67b5a9b212e615d7/F7.jpg", "media/86de67b5a9b212e615d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (silhouette of a person standing on the beach) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. forward\nC. backward\nD. up\nE. right\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "52eba7b93aa20b7e3ecc86a9f5076f7be64d69ce2a4eaf0622b611713ccf6311"}
{"qa_id": "youtube_vis__cfddad4954__camera_object_distance_change__01", "video_id": "youtube_vis__cfddad4954", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, did object A (silhouette of a person standing on the beach) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/86de67b5a9b212e615d7/F0.jpg", "media/86de67b5a9b212e615d7/F1.jpg", "media/86de67b5a9b212e615d7/F2.jpg", "media/86de67b5a9b212e615d7/F3.jpg", "media/86de67b5a9b212e615d7/F4.jpg", "media/86de67b5a9b212e615d7/F5.jpg", "media/86de67b5a9b212e615d7/F6.jpg", "media/86de67b5a9b212e615d7/F7.jpg", "media/86de67b5a9b212e615d7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (silhouette of a person standing on the beach) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8817ff6a45977af9a2d3a4b5295048dc6c11aa2aa1fdf3e4b3fa5242b1679049"}
{"qa_id": "youtube_vis__cff8191891__camera_translation_direction__00", "video_id": "youtube_vis__cff8191891", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/fcfdf4284c65c71da6a3/F0.jpg", "media/fcfdf4284c65c71da6a3/F1.jpg", "media/fcfdf4284c65c71da6a3/F2.jpg", "media/fcfdf4284c65c71da6a3/F3.jpg", "media/fcfdf4284c65c71da6a3/F4.jpg", "media/fcfdf4284c65c71da6a3/F5.jpg", "media/fcfdf4284c65c71da6a3/F6.jpg", "media/fcfdf4284c65c71da6a3/F7.jpg", "media/fcfdf4284c65c71da6a3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ebd9620a90ba465c038b323f89c88d6f94238e366d19d5c6a82dda7cd5bcd0fb"}
{"qa_id": "youtube_vis__cff8191891__object_world_motion_direction__01", "video_id": "youtube_vis__cff8191891", "task": "object_world_motion_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (white and orange front of a train approaching on the tracks) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "down", "forward", "left"], "frame_paths": ["media/dfbc802e8e6a77933ef8/F0.jpg", "media/dfbc802e8e6a77933ef8/F1.jpg", "media/dfbc802e8e6a77933ef8/F2.jpg", "media/dfbc802e8e6a77933ef8/F3.jpg", "media/dfbc802e8e6a77933ef8/F4.jpg", "media/dfbc802e8e6a77933ef8/F5.jpg", "media/dfbc802e8e6a77933ef8/F6.jpg", "media/dfbc802e8e6a77933ef8/F7.jpg", "media/dfbc802e8e6a77933ef8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (white and orange front of a train approaching on the tracks) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. down\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "811988b84415efabfc353a2d04d7ed1bf15dba945ed5484612e9d00466040795"}
{"qa_id": "youtube_vis__cff8191891__camera_object_distance_change__02", "video_id": "youtube_vis__cff8191891", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F3, did object A (white and orange front of a train approaching on the tracks) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/dfbc802e8e6a77933ef8/F0.jpg", "media/dfbc802e8e6a77933ef8/F1.jpg", "media/dfbc802e8e6a77933ef8/F2.jpg", "media/dfbc802e8e6a77933ef8/F3.jpg", "media/dfbc802e8e6a77933ef8/F4.jpg", "media/dfbc802e8e6a77933ef8/F5.jpg", "media/dfbc802e8e6a77933ef8/F6.jpg", "media/dfbc802e8e6a77933ef8/F7.jpg", "media/dfbc802e8e6a77933ef8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (white and orange front of a train approaching on the tracks) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "326396c375cdb9e3b9b1c64216de37f0d9e8b5418813c85af01c82d402c8426e"}
{"qa_id": "youtube_vis__d0b1e8b322__camera_view_direction_change__00", "video_id": "youtube_vis__d0b1e8b322", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/797347ba3158faaf9920/F0.jpg", "media/797347ba3158faaf9920/F1.jpg", "media/797347ba3158faaf9920/F2.jpg", "media/797347ba3158faaf9920/F3.jpg", "media/797347ba3158faaf9920/F4.jpg", "media/797347ba3158faaf9920/F5.jpg", "media/797347ba3158faaf9920/F6.jpg", "media/797347ba3158faaf9920/F7.jpg", "media/797347ba3158faaf9920/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "059a7f442d730e2f2310bbce22489628c393b05cb7267212c9445b9741ffabd8"}
{"qa_id": "youtube_vis__d2b026739a__camera_translation_direction__00", "video_id": "youtube_vis__d2b026739a", "task": "camera_translation_direction", "group": "HHL", "group4": "HLL+LHL+HHL", "question": "From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.", "options": ["left", "right", "up", "forward", "backward", "down"], "frame_paths": ["media/0d748b981baea56a62fb/F0.jpg", "media/0d748b981baea56a62fb/F1.jpg", "media/0d748b981baea56a62fb/F2.jpg", "media/0d748b981baea56a62fb/F3.jpg", "media/0d748b981baea56a62fb/F4.jpg", "media/0d748b981baea56a62fb/F5.jpg", "media/0d748b981baea56a62fb/F6.jpg", "media/0d748b981baea56a62fb/F7.jpg", "media/0d748b981baea56a62fb/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F5: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. right\nC. up\nD. forward\nE. backward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c3c3b1432ac45f02a778f492bbe7530f6374d3c2d88479f4b29d3f1167eb0a1"}
{"qa_id": "youtube_vis__d3b2cc407e__object_world_motion_direction__00", "video_id": "youtube_vis__d3b2cc407e", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in dark shirt riding a small wheeled vehicle) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "down", "up", "backward", "left"], "frame_paths": ["media/a300e9cf704b400a5ccf/F0.jpg", "media/a300e9cf704b400a5ccf/F1.jpg", "media/a300e9cf704b400a5ccf/F2.jpg", "media/a300e9cf704b400a5ccf/F3.jpg", "media/a300e9cf704b400a5ccf/F4.jpg", "media/a300e9cf704b400a5ccf/F5.jpg", "media/a300e9cf704b400a5ccf/F6.jpg", "media/a300e9cf704b400a5ccf/F7.jpg", "media/a300e9cf704b400a5ccf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (person in dark shirt riding a small wheeled vehicle) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. down\nD. up\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ca6cf3684249f7f6bafd32c9caf71c9b06e226ff111f8382c57c0218175f4209"}
{"qa_id": "youtube_vis__d3b2cc407e__camera_object_distance_change__01", "video_id": "youtube_vis__d3b2cc407e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F4 to F6, did object A (person in dark shirt riding a small wheeled vehicle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a300e9cf704b400a5ccf/F0.jpg", "media/a300e9cf704b400a5ccf/F1.jpg", "media/a300e9cf704b400a5ccf/F2.jpg", "media/a300e9cf704b400a5ccf/F3.jpg", "media/a300e9cf704b400a5ccf/F4.jpg", "media/a300e9cf704b400a5ccf/F5.jpg", "media/a300e9cf704b400a5ccf/F6.jpg", "media/a300e9cf704b400a5ccf/F7.jpg", "media/a300e9cf704b400a5ccf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F6, did object A (person in dark shirt riding a small wheeled vehicle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9667309fbf1cdc3d15320e3aafb3116cc4d2e9dac7bd3a43d0c9ba695ca72dcb"}
{"qa_id": "youtube_vis__d3f5c309cc__camera_translation_direction__00", "video_id": "youtube_vis__d3f5c309cc", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "left", "backward", "up", "right"], "frame_paths": ["media/b4e4325eb2d33f0f9153/F0.jpg", "media/b4e4325eb2d33f0f9153/F1.jpg", "media/b4e4325eb2d33f0f9153/F2.jpg", "media/b4e4325eb2d33f0f9153/F3.jpg", "media/b4e4325eb2d33f0f9153/F4.jpg", "media/b4e4325eb2d33f0f9153/F5.jpg", "media/b4e4325eb2d33f0f9153/F6.jpg", "media/b4e4325eb2d33f0f9153/F7.jpg", "media/b4e4325eb2d33f0f9153/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. left\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "37c11dc099d26ec7a053ec582df5ce65c92197acee0ee112aaf17145cdce78cb"}
{"qa_id": "youtube_vis__d4a607ad81__camera_view_direction_change__00", "video_id": "youtube_vis__d4a607ad81", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/ae4435bec127afdf812e/F0.jpg", "media/ae4435bec127afdf812e/F1.jpg", "media/ae4435bec127afdf812e/F2.jpg", "media/ae4435bec127afdf812e/F3.jpg", "media/ae4435bec127afdf812e/F4.jpg", "media/ae4435bec127afdf812e/F5.jpg", "media/ae4435bec127afdf812e/F6.jpg", "media/ae4435bec127afdf812e/F7.jpg", "media/ae4435bec127afdf812e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "youtube_vis__d4a607ad81__object_world_motion_direction__01", "video_id": "youtube_vis__d4a607ad81", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow with white spots on legs and body) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "backward", "right", "down", "up"], "frame_paths": ["media/8cf617ce044b6db1a643/F0.jpg", "media/8cf617ce044b6db1a643/F1.jpg", "media/8cf617ce044b6db1a643/F2.jpg", "media/8cf617ce044b6db1a643/F3.jpg", "media/8cf617ce044b6db1a643/F4.jpg", "media/8cf617ce044b6db1a643/F5.jpg", "media/8cf617ce044b6db1a643/F6.jpg", "media/8cf617ce044b6db1a643/F7.jpg", "media/8cf617ce044b6db1a643/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F8, in which dominant direction did object A (brown cow with white spots on legs and body) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. backward\nD. right\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "69ca535cda805fb96957e1f8d852f16061e860f973a8136d31bafb7c690a5db0"}
{"qa_id": "youtube_vis__dc4e812347__object_world_motion_direction__00", "video_id": "youtube_vis__dc4e812347", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (small brown dog) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "down", "backward", "right", "left", "up"], "frame_paths": ["media/38bf264ad9dc798dc3a0/F0.jpg", "media/38bf264ad9dc798dc3a0/F1.jpg", "media/38bf264ad9dc798dc3a0/F2.jpg", "media/38bf264ad9dc798dc3a0/F3.jpg", "media/38bf264ad9dc798dc3a0/F4.jpg", "media/38bf264ad9dc798dc3a0/F5.jpg", "media/38bf264ad9dc798dc3a0/F6.jpg", "media/38bf264ad9dc798dc3a0/F7.jpg", "media/38bf264ad9dc798dc3a0/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (small brown dog) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. down\nC. backward\nD. right\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fe0f93de222e605cd54fa28d2783c2fe81ae4541df280decf74b4a3bce908b1c"}
{"qa_id": "youtube_vis__dd6c46053c__camera_object_distance_change__00", "video_id": "youtube_vis__dd6c46053c", "task": "camera_object_distance_change", "group": "LLL", "group4": "LLL", "question": "From F3 to F6, did object A (tennis player in orange shirt and shorts) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/48f3c29e81294a881ce5/F0.jpg", "media/48f3c29e81294a881ce5/F1.jpg", "media/48f3c29e81294a881ce5/F2.jpg", "media/48f3c29e81294a881ce5/F3.jpg", "media/48f3c29e81294a881ce5/F4.jpg", "media/48f3c29e81294a881ce5/F5.jpg", "media/48f3c29e81294a881ce5/F6.jpg", "media/48f3c29e81294a881ce5/F7.jpg", "media/48f3c29e81294a881ce5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F6, did object A (tennis player in orange shirt and shorts) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cbe6b3fa6aa61760378dc869371c3f890bb30cc300a7c95c17787076d38fdcd7"}
{"qa_id": "youtube_vis__de2606b73a__camera_view_direction_change__00", "video_id": "youtube_vis__de2606b73a", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/79f67b4c44f9ff2c6f71/F0.jpg", "media/79f67b4c44f9ff2c6f71/F1.jpg", "media/79f67b4c44f9ff2c6f71/F2.jpg", "media/79f67b4c44f9ff2c6f71/F3.jpg", "media/79f67b4c44f9ff2c6f71/F4.jpg", "media/79f67b4c44f9ff2c6f71/F5.jpg", "media/79f67b4c44f9ff2c6f71/F6.jpg", "media/79f67b4c44f9ff2c6f71/F7.jpg", "media/79f67b4c44f9ff2c6f71/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8e3508517134a165f8ddd218d8655760426727d008d3f0a847b06c5f718e062e"}
{"qa_id": "youtube_vis__de79f45168__object_world_motion_direction__00", "video_id": "youtube_vis__de79f45168", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in red shirt and black shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "forward", "right", "down", "left", "backward"], "frame_paths": ["media/0e74affcfe55332d2523/F0.jpg", "media/0e74affcfe55332d2523/F1.jpg", "media/0e74affcfe55332d2523/F2.jpg", "media/0e74affcfe55332d2523/F3.jpg", "media/0e74affcfe55332d2523/F4.jpg", "media/0e74affcfe55332d2523/F5.jpg", "media/0e74affcfe55332d2523/F6.jpg", "media/0e74affcfe55332d2523/F7.jpg", "media/0e74affcfe55332d2523/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F3, in which dominant direction did object B (person in red shirt and black shorts) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. forward\nC. right\nD. down\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "cdf0de8b51072c580c0984f72c06da0d4b0f14d470cc3844a10c4e93cf5af4ec"}
{"qa_id": "youtube_vis__df3d268e87__camera_translation_direction__00", "video_id": "youtube_vis__df3d268e87", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "left", "down", "backward", "up"], "frame_paths": ["media/5f3fc36a6880c6673ef5/F0.jpg", "media/5f3fc36a6880c6673ef5/F1.jpg", "media/5f3fc36a6880c6673ef5/F2.jpg", "media/5f3fc36a6880c6673ef5/F3.jpg", "media/5f3fc36a6880c6673ef5/F4.jpg", "media/5f3fc36a6880c6673ef5/F5.jpg", "media/5f3fc36a6880c6673ef5/F6.jpg", "media/5f3fc36a6880c6673ef5/F7.jpg", "media/5f3fc36a6880c6673ef5/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. left\nD. down\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "96a3a1b298a30c3a213c3a4ac7ab99739e0cd56db6fdf1ab971d6c854560a8c8"}
{"qa_id": "youtube_vis__e0de82caa7__camera_translation_direction__00", "video_id": "youtube_vis__e0de82caa7", "task": "camera_translation_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "down", "forward", "right", "left", "backward"], "frame_paths": ["media/1643b42db986d3afe277/F0.jpg", "media/1643b42db986d3afe277/F1.jpg", "media/1643b42db986d3afe277/F2.jpg", "media/1643b42db986d3afe277/F3.jpg", "media/1643b42db986d3afe277/F4.jpg", "media/1643b42db986d3afe277/F5.jpg", "media/1643b42db986d3afe277/F6.jpg", "media/1643b42db986d3afe277/F7.jpg", "media/1643b42db986d3afe277/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. forward\nD. right\nE. left\nF. backward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4395c07280c14a4e88c514d530424db2f6a1ffc729a9b9e9b5e34ac6325eb4df"}
{"qa_id": "youtube_vis__e0de82caa7__camera_view_direction_change__01", "video_id": "youtube_vis__e0de82caa7", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?", "options": ["right", "up", "left", "down"], "frame_paths": ["media/1643b42db986d3afe277/F0.jpg", "media/1643b42db986d3afe277/F1.jpg", "media/1643b42db986d3afe277/F2.jpg", "media/1643b42db986d3afe277/F3.jpg", "media/1643b42db986d3afe277/F4.jpg", "media/1643b42db986d3afe277/F5.jpg", "media/1643b42db986d3afe277/F6.jpg", "media/1643b42db986d3afe277/F7.jpg", "media/1643b42db986d3afe277/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F6?\n\nOptions:\nA. right\nB. up\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2125aef20f2f9e771f5a10d7239444f943f23d009b407007224a3add69a33bf2"}
{"qa_id": "youtube_vis__e0de82caa7__object_world_motion_direction__02", "video_id": "youtube_vis__e0de82caa7", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (giant panda with black and white fur) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "left", "backward", "forward"], "frame_paths": ["media/0aa5238cae2e59953006/F0.jpg", "media/0aa5238cae2e59953006/F1.jpg", "media/0aa5238cae2e59953006/F2.jpg", "media/0aa5238cae2e59953006/F3.jpg", "media/0aa5238cae2e59953006/F4.jpg", "media/0aa5238cae2e59953006/F5.jpg", "media/0aa5238cae2e59953006/F6.jpg", "media/0aa5238cae2e59953006/F7.jpg", "media/0aa5238cae2e59953006/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (giant panda with black and white fur) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. left\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "85aea3e001f9e7603ed2b01bc4cb5604d24e0e601f224eb944f6ce7e73afefea"}
{"qa_id": "youtube_vis__e1eff80c9a__camera_view_direction_change__00", "video_id": "youtube_vis__e1eff80c9a", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/7f59c9c9368dbc68421d/F0.jpg", "media/7f59c9c9368dbc68421d/F1.jpg", "media/7f59c9c9368dbc68421d/F2.jpg", "media/7f59c9c9368dbc68421d/F3.jpg", "media/7f59c9c9368dbc68421d/F4.jpg", "media/7f59c9c9368dbc68421d/F5.jpg", "media/7f59c9c9368dbc68421d/F6.jpg", "media/7f59c9c9368dbc68421d/F7.jpg", "media/7f59c9c9368dbc68421d/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fed7f52831e0bcd8eb5690bd6b9d90fabd761945524b757e41c75e8a5483c227"}
{"qa_id": "youtube_vis__e1eff80c9a__object_world_motion_direction__01", "video_id": "youtube_vis__e1eff80c9a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (man in dark blue jacket and conductor hat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "left", "forward", "down"], "frame_paths": ["media/9685ee6f41b3d0839a57/F0.jpg", "media/9685ee6f41b3d0839a57/F1.jpg", "media/9685ee6f41b3d0839a57/F2.jpg", "media/9685ee6f41b3d0839a57/F3.jpg", "media/9685ee6f41b3d0839a57/F4.jpg", "media/9685ee6f41b3d0839a57/F5.jpg", "media/9685ee6f41b3d0839a57/F6.jpg", "media/9685ee6f41b3d0839a57/F7.jpg", "media/9685ee6f41b3d0839a57/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (man in dark blue jacket and conductor hat) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1fbb7b401b36e69cde7e2a0179a1e11c9a3a42416f8eb4689267b08c4f69ff36"}
{"qa_id": "youtube_vis__e1eff80c9a__camera_object_distance_change__02", "video_id": "youtube_vis__e1eff80c9a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (man in dark blue jacket and conductor hat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9685ee6f41b3d0839a57/F0.jpg", "media/9685ee6f41b3d0839a57/F1.jpg", "media/9685ee6f41b3d0839a57/F2.jpg", "media/9685ee6f41b3d0839a57/F3.jpg", "media/9685ee6f41b3d0839a57/F4.jpg", "media/9685ee6f41b3d0839a57/F5.jpg", "media/9685ee6f41b3d0839a57/F6.jpg", "media/9685ee6f41b3d0839a57/F7.jpg", "media/9685ee6f41b3d0839a57/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (man in dark blue jacket and conductor hat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b7c5be3a46526e47f016430bb9b304e98b5e11662fa8bc7f179398c175fc30a5"}
{"qa_id": "youtube_vis__e1eff80c9a__object_world_motion_direction__03", "video_id": "youtube_vis__e1eff80c9a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object B (small red toy train engine) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "down", "left", "up"], "frame_paths": ["media/5c6629e26ca001535cad/F0.jpg", "media/5c6629e26ca001535cad/F1.jpg", "media/5c6629e26ca001535cad/F2.jpg", "media/5c6629e26ca001535cad/F3.jpg", "media/5c6629e26ca001535cad/F4.jpg", "media/5c6629e26ca001535cad/F5.jpg", "media/5c6629e26ca001535cad/F6.jpg", "media/5c6629e26ca001535cad/F7.jpg", "media/5c6629e26ca001535cad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object B (small red toy train engine) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. down\nE. left\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ba50ad1e0f5e340ea2f15074fe516fbbaa26c265bc5e0114c9eac2fc8f87bf43"}
{"qa_id": "youtube_vis__e1eff80c9a__camera_object_distance_change__04", "video_id": "youtube_vis__e1eff80c9a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object B (small red toy train engine) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/5c6629e26ca001535cad/F0.jpg", "media/5c6629e26ca001535cad/F1.jpg", "media/5c6629e26ca001535cad/F2.jpg", "media/5c6629e26ca001535cad/F3.jpg", "media/5c6629e26ca001535cad/F4.jpg", "media/5c6629e26ca001535cad/F5.jpg", "media/5c6629e26ca001535cad/F6.jpg", "media/5c6629e26ca001535cad/F7.jpg", "media/5c6629e26ca001535cad/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object B (small red toy train engine) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f6740e20f109ec5ebd6f8f4094b558b1c4c7d6917b5bf774c0d8066d5aab6174"}
{"qa_id": "youtube_vis__e2ccffb890__camera_view_direction_change__00", "video_id": "youtube_vis__e2ccffb890", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/02a474f0d5da47ee8f8f/F0.jpg", "media/02a474f0d5da47ee8f8f/F1.jpg", "media/02a474f0d5da47ee8f8f/F2.jpg", "media/02a474f0d5da47ee8f8f/F3.jpg", "media/02a474f0d5da47ee8f8f/F4.jpg", "media/02a474f0d5da47ee8f8f/F5.jpg", "media/02a474f0d5da47ee8f8f/F6.jpg", "media/02a474f0d5da47ee8f8f/F7.jpg", "media/02a474f0d5da47ee8f8f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "badcf446ceefa9b7ba2b3a7428b85742f2a82b7fa5cf190f9e2255ec052110c7"}
{"qa_id": "youtube_vis__e2ea25542c__camera_translation_direction__00", "video_id": "youtube_vis__e2ea25542c", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["forward", "backward", "left", "down", "up", "right"], "frame_paths": ["media/efe67b72d220a647d670/F0.jpg", "media/efe67b72d220a647d670/F1.jpg", "media/efe67b72d220a647d670/F2.jpg", "media/efe67b72d220a647d670/F3.jpg", "media/efe67b72d220a647d670/F4.jpg", "media/efe67b72d220a647d670/F5.jpg", "media/efe67b72d220a647d670/F6.jpg", "media/efe67b72d220a647d670/F7.jpg", "media/efe67b72d220a647d670/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. backward\nC. left\nD. down\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "80622f5c0b0566e73920da8f1ffb1583cca4a647c8ef0dbb295cc4a22391e604"}
{"qa_id": "youtube_vis__e2ea25542c__camera_object_distance_change__01", "video_id": "youtube_vis__e2ea25542c", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F4 to F8, did object A (small mottled fish with a dark tail) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/a3aa0b66493c4404c6fc/F0.jpg", "media/a3aa0b66493c4404c6fc/F1.jpg", "media/a3aa0b66493c4404c6fc/F2.jpg", "media/a3aa0b66493c4404c6fc/F3.jpg", "media/a3aa0b66493c4404c6fc/F4.jpg", "media/a3aa0b66493c4404c6fc/F5.jpg", "media/a3aa0b66493c4404c6fc/F6.jpg", "media/a3aa0b66493c4404c6fc/F7.jpg", "media/a3aa0b66493c4404c6fc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F4 to F8, did object A (small mottled fish with a dark tail) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "523dbfde0f6ae88805707c49e3987971ed656a00c47a92cdab106fdbf988e946"}
{"qa_id": "youtube_vis__e3ee20ce97__object_world_motion_direction__00", "video_id": "youtube_vis__e3ee20ce97", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F4, in which dominant direction did object A (small tan dog) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "left", "down", "backward", "up", "right"], "frame_paths": ["media/467774421d1212f41226/F0.jpg", "media/467774421d1212f41226/F1.jpg", "media/467774421d1212f41226/F2.jpg", "media/467774421d1212f41226/F3.jpg", "media/467774421d1212f41226/F4.jpg", "media/467774421d1212f41226/F5.jpg", "media/467774421d1212f41226/F6.jpg", "media/467774421d1212f41226/F7.jpg", "media/467774421d1212f41226/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F4, in which dominant direction did object A (small tan dog) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. left\nC. down\nD. backward\nE. up\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "179687e5bd0b8864fdd5ea34f0ed4db509b6458366f1920d2db2918a6c509a22"}
{"qa_id": "youtube_vis__e3ee20ce97__camera_object_distance_change__01", "video_id": "youtube_vis__e3ee20ce97", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did object A (small tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/467774421d1212f41226/F0.jpg", "media/467774421d1212f41226/F1.jpg", "media/467774421d1212f41226/F2.jpg", "media/467774421d1212f41226/F3.jpg", "media/467774421d1212f41226/F4.jpg", "media/467774421d1212f41226/F5.jpg", "media/467774421d1212f41226/F6.jpg", "media/467774421d1212f41226/F7.jpg", "media/467774421d1212f41226/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did object A (small tan dog) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "708595ebfa5a925e1dc9e35f51878e669c84a423d77bd1672674ac872d9a1cdc"}
{"qa_id": "youtube_vis__e3ee20ce97__object_pair_distance_change__02", "video_id": "youtube_vis__e3ee20ce97", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F4, did the 3D distance between object A (small tan dog) and object B (person in white shirt) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/09d1f858bfbd0017a7e4/F0.jpg", "media/09d1f858bfbd0017a7e4/F1.jpg", "media/09d1f858bfbd0017a7e4/F2.jpg", "media/09d1f858bfbd0017a7e4/F3.jpg", "media/09d1f858bfbd0017a7e4/F4.jpg", "media/09d1f858bfbd0017a7e4/F5.jpg", "media/09d1f858bfbd0017a7e4/F6.jpg", "media/09d1f858bfbd0017a7e4/F7.jpg", "media/09d1f858bfbd0017a7e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, did the 3D distance between object A (small tan dog) and object B (person in white shirt) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "eaf8bb72839830c4ce8f3a895af1ddb189c506ae3cbdec46bd6804fa5c6de97c"}
{"qa_id": "youtube_vis__e3ee20ce97__object_travel_distance_comparison__03", "video_id": "youtube_vis__e3ee20ce97", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small tan dog) or B (person in white shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/09d1f858bfbd0017a7e4/F0.jpg", "media/09d1f858bfbd0017a7e4/F1.jpg", "media/09d1f858bfbd0017a7e4/F2.jpg", "media/09d1f858bfbd0017a7e4/F3.jpg", "media/09d1f858bfbd0017a7e4/F4.jpg", "media/09d1f858bfbd0017a7e4/F5.jpg", "media/09d1f858bfbd0017a7e4/F6.jpg", "media/09d1f858bfbd0017a7e4/F7.jpg", "media/09d1f858bfbd0017a7e4/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2", "g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F4, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (small tan dog) or B (person in white shirt)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "85366af85affd13e04adfac9cb81778e58cf1d02d8ed1fa5f4782b72df3bf9ce"}
{"qa_id": "youtube_vis__e4b2095f58__camera_object_distance_change__00", "video_id": "youtube_vis__e4b2095f58", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F1 to F4, did object A (tan dog with pointed ears walking on a gravel road) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/3aa95ca25f48bffc6b7e/F0.jpg", "media/3aa95ca25f48bffc6b7e/F1.jpg", "media/3aa95ca25f48bffc6b7e/F2.jpg", "media/3aa95ca25f48bffc6b7e/F3.jpg", "media/3aa95ca25f48bffc6b7e/F4.jpg", "media/3aa95ca25f48bffc6b7e/F5.jpg", "media/3aa95ca25f48bffc6b7e/F6.jpg", "media/3aa95ca25f48bffc6b7e/F7.jpg", "media/3aa95ca25f48bffc6b7e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F1 to F4, did object A (tan dog with pointed ears walking on a gravel road) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8a8598f09502e561d1e069078ad44fbb455623511efd43b25e6c0704baac274b"}
{"qa_id": "youtube_vis__e979fce0d2__camera_view_direction_change__00", "video_id": "youtube_vis__e979fce0d2", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "down", "up", "right"], "frame_paths": ["media/f8c70633abc623c4d5a3/F0.jpg", "media/f8c70633abc623c4d5a3/F1.jpg", "media/f8c70633abc623c4d5a3/F2.jpg", "media/f8c70633abc623c4d5a3/F3.jpg", "media/f8c70633abc623c4d5a3/F4.jpg", "media/f8c70633abc623c4d5a3/F5.jpg", "media/f8c70633abc623c4d5a3/F6.jpg", "media/f8c70633abc623c4d5a3/F7.jpg", "media/f8c70633abc623c4d5a3/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. down\nC. up\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7c14dcd5f24642225f3bd8464554491ae5cd58ca8706517ba6816f0d0ae533c8"}
{"qa_id": "youtube_vis__e979fce0d2__object_world_motion_direction__01", "video_id": "youtube_vis__e979fce0d2", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (white rally car with black and orange graphics) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "right", "forward", "backward", "down", "left"], "frame_paths": ["media/d3af09b1d0a379c5bdfc/F0.jpg", "media/d3af09b1d0a379c5bdfc/F1.jpg", "media/d3af09b1d0a379c5bdfc/F2.jpg", "media/d3af09b1d0a379c5bdfc/F3.jpg", "media/d3af09b1d0a379c5bdfc/F4.jpg", "media/d3af09b1d0a379c5bdfc/F5.jpg", "media/d3af09b1d0a379c5bdfc/F6.jpg", "media/d3af09b1d0a379c5bdfc/F7.jpg", "media/d3af09b1d0a379c5bdfc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (white rally car with black and orange graphics) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. right\nC. forward\nD. backward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b9e97b78756b7a03d14c87249ba74c1a0d1c88b3bf2d59623f6ba3d4c707ab4b"}
{"qa_id": "youtube_vis__e979fce0d2__camera_object_distance_change__02", "video_id": "youtube_vis__e979fce0d2", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (white rally car with black and orange graphics) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d3af09b1d0a379c5bdfc/F0.jpg", "media/d3af09b1d0a379c5bdfc/F1.jpg", "media/d3af09b1d0a379c5bdfc/F2.jpg", "media/d3af09b1d0a379c5bdfc/F3.jpg", "media/d3af09b1d0a379c5bdfc/F4.jpg", "media/d3af09b1d0a379c5bdfc/F5.jpg", "media/d3af09b1d0a379c5bdfc/F6.jpg", "media/d3af09b1d0a379c5bdfc/F7.jpg", "media/d3af09b1d0a379c5bdfc/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (white rally car with black and orange graphics) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "f446fd69d67c8a66db47bb025a4e50f1d91e97c4be99573b2ca74a801332e478"}
{"qa_id": "youtube_vis__ea04120098__object_world_motion_direction__00", "video_id": "youtube_vis__ea04120098", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a white helmet and dark shirt riding a scooter) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "forward", "right", "up", "down"], "frame_paths": ["media/9a5efc3952866e6d7803/F0.jpg", "media/9a5efc3952866e6d7803/F1.jpg", "media/9a5efc3952866e6d7803/F2.jpg", "media/9a5efc3952866e6d7803/F3.jpg", "media/9a5efc3952866e6d7803/F4.jpg", "media/9a5efc3952866e6d7803/F5.jpg", "media/9a5efc3952866e6d7803/F6.jpg", "media/9a5efc3952866e6d7803/F7.jpg", "media/9a5efc3952866e6d7803/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (person wearing a white helmet and dark shirt riding a scooter) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. forward\nD. right\nE. up\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "32ea89832ff2e69e631c011d7585b2362020b9f0658a8398157e4a49eccd56e9"}
{"qa_id": "youtube_vis__ea04120098__camera_object_distance_change__01", "video_id": "youtube_vis__ea04120098", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (person wearing a white helmet and dark shirt riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/9a5efc3952866e6d7803/F0.jpg", "media/9a5efc3952866e6d7803/F1.jpg", "media/9a5efc3952866e6d7803/F2.jpg", "media/9a5efc3952866e6d7803/F3.jpg", "media/9a5efc3952866e6d7803/F4.jpg", "media/9a5efc3952866e6d7803/F5.jpg", "media/9a5efc3952866e6d7803/F6.jpg", "media/9a5efc3952866e6d7803/F7.jpg", "media/9a5efc3952866e6d7803/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (person wearing a white helmet and dark shirt riding a scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4c0aa5174199efd5fc2e54d9d187ee3a11dcbd829209dbf1da3cfd83a1e37a1c"}
{"qa_id": "youtube_vis__ea04120098__object_world_motion_direction__02", "video_id": "youtube_vis__ea04120098", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object B (cream and green motor scooter) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "forward", "up"], "frame_paths": ["media/26162c19d4ad9a21cd1c/F0.jpg", "media/26162c19d4ad9a21cd1c/F1.jpg", "media/26162c19d4ad9a21cd1c/F2.jpg", "media/26162c19d4ad9a21cd1c/F3.jpg", "media/26162c19d4ad9a21cd1c/F4.jpg", "media/26162c19d4ad9a21cd1c/F5.jpg", "media/26162c19d4ad9a21cd1c/F6.jpg", "media/26162c19d4ad9a21cd1c/F7.jpg", "media/26162c19d4ad9a21cd1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object B (cream and green motor scooter) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ab8304b212c0621a6d5e3b752c2018fd9265f81d697fc16419fe8fd2d14ab3b6"}
{"qa_id": "youtube_vis__ea04120098__camera_object_distance_change__03", "video_id": "youtube_vis__ea04120098", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object B (cream and green motor scooter) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/26162c19d4ad9a21cd1c/F0.jpg", "media/26162c19d4ad9a21cd1c/F1.jpg", "media/26162c19d4ad9a21cd1c/F2.jpg", "media/26162c19d4ad9a21cd1c/F3.jpg", "media/26162c19d4ad9a21cd1c/F4.jpg", "media/26162c19d4ad9a21cd1c/F5.jpg", "media/26162c19d4ad9a21cd1c/F6.jpg", "media/26162c19d4ad9a21cd1c/F7.jpg", "media/26162c19d4ad9a21cd1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object B (cream and green motor scooter) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "ee6644d5673747d7cc8d4894a3e53eb8e5a22361c3bfd27462febfe1585a3a6f"}
{"qa_id": "youtube_vis__ed8d7e7513__camera_translation_direction__00", "video_id": "youtube_vis__ed8d7e7513", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "down", "left", "backward", "forward", "up"], "frame_paths": ["media/00b33b987b1841ec5244/F0.jpg", "media/00b33b987b1841ec5244/F1.jpg", "media/00b33b987b1841ec5244/F2.jpg", "media/00b33b987b1841ec5244/F3.jpg", "media/00b33b987b1841ec5244/F4.jpg", "media/00b33b987b1841ec5244/F5.jpg", "media/00b33b987b1841ec5244/F6.jpg", "media/00b33b987b1841ec5244/F7.jpg", "media/00b33b987b1841ec5244/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, in which dominant direction did the camera center move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. down\nC. left\nD. backward\nE. forward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "38a7c9feb8d72642d231882895094c91d36df0de0123873d22030676e44c2287"}
{"qa_id": "youtube_vis__ed8d7e7513__object_world_motion_direction__01", "video_id": "youtube_vis__ed8d7e7513", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F7, in which dominant direction did object A (snowboarder in orange pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "down", "left", "up", "forward"], "frame_paths": ["media/b120751abe8a123d8cc6/F0.jpg", "media/b120751abe8a123d8cc6/F1.jpg", "media/b120751abe8a123d8cc6/F2.jpg", "media/b120751abe8a123d8cc6/F3.jpg", "media/b120751abe8a123d8cc6/F4.jpg", "media/b120751abe8a123d8cc6/F5.jpg", "media/b120751abe8a123d8cc6/F6.jpg", "media/b120751abe8a123d8cc6/F7.jpg", "media/b120751abe8a123d8cc6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F7, in which dominant direction did object A (snowboarder in orange pants) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. down\nD. left\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "8fe652f84349a8fa70ce2919247ac48750b71c700fd0c594c8663c594a23cfcd"}
{"qa_id": "youtube_vis__ed8d7e7513__camera_object_distance_change__02", "video_id": "youtube_vis__ed8d7e7513", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F7, did object A (snowboarder in orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/b120751abe8a123d8cc6/F0.jpg", "media/b120751abe8a123d8cc6/F1.jpg", "media/b120751abe8a123d8cc6/F2.jpg", "media/b120751abe8a123d8cc6/F3.jpg", "media/b120751abe8a123d8cc6/F4.jpg", "media/b120751abe8a123d8cc6/F5.jpg", "media/b120751abe8a123d8cc6/F6.jpg", "media/b120751abe8a123d8cc6/F7.jpg", "media/b120751abe8a123d8cc6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F7, did object A (snowboarder in orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "fb0867f1a778893378c5ee3db3bc7676f7883bc52e8c447c61e80e35ac759baf"}
{"qa_id": "youtube_vis__edcf78a43b__camera_translation_direction__00", "video_id": "youtube_vis__edcf78a43b", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["left", "backward", "down", "right", "up", "forward"], "frame_paths": ["media/6bb7f6d4e45e1625c65e/F0.jpg", "media/6bb7f6d4e45e1625c65e/F1.jpg", "media/6bb7f6d4e45e1625c65e/F2.jpg", "media/6bb7f6d4e45e1625c65e/F3.jpg", "media/6bb7f6d4e45e1625c65e/F4.jpg", "media/6bb7f6d4e45e1625c65e/F5.jpg", "media/6bb7f6d4e45e1625c65e/F6.jpg", "media/6bb7f6d4e45e1625c65e/F7.jpg", "media/6bb7f6d4e45e1625c65e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. backward\nC. down\nD. right\nE. up\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6d49d062574472099af50e383a9519f5765c1bb4c2e8bea7777a023d58945fcc"}
{"qa_id": "youtube_vis__edcf78a43b__camera_view_direction_change__01", "video_id": "youtube_vis__edcf78a43b", "task": "camera_view_direction_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["up", "right", "down", "left"], "frame_paths": ["media/6bb7f6d4e45e1625c65e/F0.jpg", "media/6bb7f6d4e45e1625c65e/F1.jpg", "media/6bb7f6d4e45e1625c65e/F2.jpg", "media/6bb7f6d4e45e1625c65e/F3.jpg", "media/6bb7f6d4e45e1625c65e/F4.jpg", "media/6bb7f6d4e45e1625c65e/F5.jpg", "media/6bb7f6d4e45e1625c65e/F6.jpg", "media/6bb7f6d4e45e1625c65e/F7.jpg", "media/6bb7f6d4e45e1625c65e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. up\nB. right\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "480e55fd8a65ac9933e67a2d8f3138ca5da9e9879630fe1c30a3d92892fcd8ff"}
{"qa_id": "youtube_vis__edcf78a43b__object_world_motion_direction__02", "video_id": "youtube_vis__edcf78a43b", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object B (dark brown horse) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["backward", "up", "right", "left", "forward", "down"], "frame_paths": ["media/7459bfc20bb96247ee6f/F0.jpg", "media/7459bfc20bb96247ee6f/F1.jpg", "media/7459bfc20bb96247ee6f/F2.jpg", "media/7459bfc20bb96247ee6f/F3.jpg", "media/7459bfc20bb96247ee6f/F4.jpg", "media/7459bfc20bb96247ee6f/F5.jpg", "media/7459bfc20bb96247ee6f/F6.jpg", "media/7459bfc20bb96247ee6f/F7.jpg", "media/7459bfc20bb96247ee6f/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object B (dark brown horse) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. up\nC. right\nD. left\nE. forward\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "965ab54158110209ba0cef94215142bd131c89571610776fc38d6023e22d6c90"}
{"qa_id": "youtube_vis__edd663afa3__camera_view_direction_change__00", "video_id": "youtube_vis__edd663afa3", "task": "camera_view_direction_change", "group": "LHL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/f8cc13d1a0fe99ad7d8e/F0.jpg", "media/f8cc13d1a0fe99ad7d8e/F1.jpg", "media/f8cc13d1a0fe99ad7d8e/F2.jpg", "media/f8cc13d1a0fe99ad7d8e/F3.jpg", "media/f8cc13d1a0fe99ad7d8e/F4.jpg", "media/f8cc13d1a0fe99ad7d8e/F5.jpg", "media/f8cc13d1a0fe99ad7d8e/F6.jpg", "media/f8cc13d1a0fe99ad7d8e/F7.jpg", "media/f8cc13d1a0fe99ad7d8e/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "7cb4feef62c3d9efd6762a749f2132199fae31e2272a2ff8d2b33844b5eac135"}
{"qa_id": "youtube_vis__ef6c2a2fb0__camera_view_direction_change__00", "video_id": "youtube_vis__ef6c2a2fb0", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?", "options": ["right", "left", "up", "down"], "frame_paths": ["media/1356b68a1147a96e5f18/F0.jpg", "media/1356b68a1147a96e5f18/F1.jpg", "media/1356b68a1147a96e5f18/F2.jpg", "media/1356b68a1147a96e5f18/F3.jpg", "media/1356b68a1147a96e5f18/F4.jpg", "media/1356b68a1147a96e5f18/F5.jpg", "media/1356b68a1147a96e5f18/F6.jpg", "media/1356b68a1147a96e5f18/F7.jpg", "media/1356b68a1147a96e5f18/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, toward which direction did the camera optical axis turn, using the camera axes frozen at F3?\n\nOptions:\nA. right\nB. left\nC. up\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "79a144e54bd2cb97522c8bd091b9cc5e6a32f8092390802b208e732a34f9748a"}
{"qa_id": "youtube_vis__ef96501dd0__object_world_motion_direction__00", "video_id": "youtube_vis__ef96501dd0", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (copper-colored cougar walking on a fallen tree branch) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "left", "down", "backward", "right"], "frame_paths": ["media/5f1730d382572e4832bf/F0.jpg", "media/5f1730d382572e4832bf/F1.jpg", "media/5f1730d382572e4832bf/F2.jpg", "media/5f1730d382572e4832bf/F3.jpg", "media/5f1730d382572e4832bf/F4.jpg", "media/5f1730d382572e4832bf/F5.jpg", "media/5f1730d382572e4832bf/F6.jpg", "media/5f1730d382572e4832bf/F7.jpg", "media/5f1730d382572e4832bf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (copper-colored cougar walking on a fallen tree branch) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. left\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6de2b82ba38c38c43683151f56289dae6fe6b2e56325507c19b4942c64c483a9"}
{"qa_id": "youtube_vis__f272aea880__object_world_motion_direction__00", "video_id": "youtube_vis__f272aea880", "task": "object_world_motion_direction", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F2, in which dominant direction did object A (red hatchback car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["forward", "up", "left", "down", "backward", "right"], "frame_paths": ["media/d604413b8d662cdd5808/F0.jpg", "media/d604413b8d662cdd5808/F1.jpg", "media/d604413b8d662cdd5808/F2.jpg", "media/d604413b8d662cdd5808/F3.jpg", "media/d604413b8d662cdd5808/F4.jpg", "media/d604413b8d662cdd5808/F5.jpg", "media/d604413b8d662cdd5808/F6.jpg", "media/d604413b8d662cdd5808/F7.jpg", "media/d604413b8d662cdd5808/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F2, in which dominant direction did object A (red hatchback car) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. up\nC. left\nD. down\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "2063433021f40aa4c906132605858f5abae2c1008564e3d3c85acf1ea022d666"}
{"qa_id": "youtube_vis__f272aea880__camera_object_distance_change__01", "video_id": "youtube_vis__f272aea880", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F2, did object A (red hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/d604413b8d662cdd5808/F0.jpg", "media/d604413b8d662cdd5808/F1.jpg", "media/d604413b8d662cdd5808/F2.jpg", "media/d604413b8d662cdd5808/F3.jpg", "media/d604413b8d662cdd5808/F4.jpg", "media/d604413b8d662cdd5808/F5.jpg", "media/d604413b8d662cdd5808/F6.jpg", "media/d604413b8d662cdd5808/F7.jpg", "media/d604413b8d662cdd5808/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (red hatchback car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "287e09939bae8c7557a824d1caf5bef03028cc44e76f62300625fb187b519920"}
{"qa_id": "youtube_vis__f38e5aa5b4__object_world_motion_direction__00", "video_id": "youtube_vis__f38e5aa5b4", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F5, in which dominant direction did object A (white city bus with number 1303 and route M1) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "up", "down", "forward", "backward", "left"], "frame_paths": ["media/a59a3eec05dc47d1df89/F0.jpg", "media/a59a3eec05dc47d1df89/F1.jpg", "media/a59a3eec05dc47d1df89/F2.jpg", "media/a59a3eec05dc47d1df89/F3.jpg", "media/a59a3eec05dc47d1df89/F4.jpg", "media/a59a3eec05dc47d1df89/F5.jpg", "media/a59a3eec05dc47d1df89/F6.jpg", "media/a59a3eec05dc47d1df89/F7.jpg", "media/a59a3eec05dc47d1df89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F5, in which dominant direction did object A (white city bus with number 1303 and route M1) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. up\nC. down\nD. forward\nE. backward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "3ed3043202c756f213b938d2e073355f29f64f84f11ecfebeef8ac0da2fb4502"}
{"qa_id": "youtube_vis__f38e5aa5b4__camera_object_distance_change__01", "video_id": "youtube_vis__f38e5aa5b4", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did object A (white city bus with number 1303 and route M1) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/a59a3eec05dc47d1df89/F0.jpg", "media/a59a3eec05dc47d1df89/F1.jpg", "media/a59a3eec05dc47d1df89/F2.jpg", "media/a59a3eec05dc47d1df89/F3.jpg", "media/a59a3eec05dc47d1df89/F4.jpg", "media/a59a3eec05dc47d1df89/F5.jpg", "media/a59a3eec05dc47d1df89/F6.jpg", "media/a59a3eec05dc47d1df89/F7.jpg", "media/a59a3eec05dc47d1df89/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (white city bus with number 1303 and route M1) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "d9c8ff2426a139f6be60c0d40d8e98d26765c6fda628f2e35c7ec847fc8fe6c7"}
{"qa_id": "youtube_vis__f38e5aa5b4__object_pair_distance_change__02", "video_id": "youtube_vis__f38e5aa5b4", "task": "object_pair_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F5, did the 3D distance between object A (white city bus with number 1303 and route M1) and object B (person in dark jacket and green pants) increase or decrease? Compare their separation at the two specified frames.", "options": ["decreased", "increased"], "frame_paths": ["media/71738ebb766b6c8c80e8/F0.jpg", "media/71738ebb766b6c8c80e8/F1.jpg", "media/71738ebb766b6c8c80e8/F2.jpg", "media/71738ebb766b6c8c80e8/F3.jpg", "media/71738ebb766b6c8c80e8/F4.jpg", "media/71738ebb766b6c8c80e8/F5.jpg", "media/71738ebb766b6c8c80e8/F6.jpg", "media/71738ebb766b6c8c80e8/F7.jpg", "media/71738ebb766b6c8c80e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did the 3D distance between object A (white city bus with number 1303 and route M1) and object B (person in dark jacket and green pants) increase or decrease? Compare their separation at the two specified frames.\n\nOptions:\nA. decreased\nB. increased\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "88c19553f9e0284f84f9869c542052a2e6280b470a2ea53617aad9273efafead"}
{"qa_id": "youtube_vis__f38e5aa5b4__object_travel_distance_comparison__03", "video_id": "youtube_vis__f38e5aa5b4", "task": "object_travel_distance_comparison", "group": "LLH", "group4": "LLH", "question": "Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white city bus with number 1303 and route M1) or B (person in dark jacket and green pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.", "options": ["B", "A"], "frame_paths": ["media/71738ebb766b6c8c80e8/F0.jpg", "media/71738ebb766b6c8c80e8/F1.jpg", "media/71738ebb766b6c8c80e8/F2.jpg", "media/71738ebb766b6c8c80e8/F3.jpg", "media/71738ebb766b6c8c80e8/F4.jpg", "media/71738ebb766b6c8c80e8/F5.jpg", "media/71738ebb766b6c8c80e8/F6.jpg", "media/71738ebb766b6c8c80e8/F7.jpg", "media/71738ebb766b6c8c80e8/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1", "g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: Over the sampled trajectory from F0 through F5, which object traveled a greater cumulative 3D distance after compensating for camera motion: A (white city bus with number 1303 and route M1) or B (person in dark jacket and green pants)? Sum consecutive sampled 3D displacements, rather than using only the start-to-end displacement.\n\nOptions:\nA. Object B\nB. Object A\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "e666bf7be659ca8b6259fc0f6e66c107685b58ef2791a1dfc4e1778997a91e4d"}
{"qa_id": "youtube_vis__f3b5e6c84a__object_world_motion_direction__00", "video_id": "youtube_vis__f3b5e6c84a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object A (dark car with bright headlights on the left side of the track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "forward", "left"], "frame_paths": ["media/dff6e2529533dae99140/F0.jpg", "media/dff6e2529533dae99140/F1.jpg", "media/dff6e2529533dae99140/F2.jpg", "media/dff6e2529533dae99140/F3.jpg", "media/dff6e2529533dae99140/F4.jpg", "media/dff6e2529533dae99140/F5.jpg", "media/dff6e2529533dae99140/F6.jpg", "media/dff6e2529533dae99140/F7.jpg", "media/dff6e2529533dae99140/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object A (dark car with bright headlights on the left side of the track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. forward\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0471b2edb9ca0bf4ce853c36a61477d6c6e6cdcb784174525608e50df5330402"}
{"qa_id": "youtube_vis__f3b5e6c84a__camera_object_distance_change__01", "video_id": "youtube_vis__f3b5e6c84a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (dark car with bright headlights on the left side of the track) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/dff6e2529533dae99140/F0.jpg", "media/dff6e2529533dae99140/F1.jpg", "media/dff6e2529533dae99140/F2.jpg", "media/dff6e2529533dae99140/F3.jpg", "media/dff6e2529533dae99140/F4.jpg", "media/dff6e2529533dae99140/F5.jpg", "media/dff6e2529533dae99140/F6.jpg", "media/dff6e2529533dae99140/F7.jpg", "media/dff6e2529533dae99140/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (dark car with bright headlights on the left side of the track) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "41e98a824925a4b91b7a3ec552dc6c381d74c0076538d6c72747ad6cda5954d3"}
{"qa_id": "youtube_vis__f3b5e6c84a__object_world_motion_direction__02", "video_id": "youtube_vis__f3b5e6c84a", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F0 to F6, in which dominant direction did object B (dark car with bright headlights on the right side of the track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "backward", "up", "right", "down"], "frame_paths": ["media/db9fda5637b803d752ba/F0.jpg", "media/db9fda5637b803d752ba/F1.jpg", "media/db9fda5637b803d752ba/F2.jpg", "media/db9fda5637b803d752ba/F3.jpg", "media/db9fda5637b803d752ba/F4.jpg", "media/db9fda5637b803d752ba/F5.jpg", "media/db9fda5637b803d752ba/F6.jpg", "media/db9fda5637b803d752ba/F7.jpg", "media/db9fda5637b803d752ba/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F6, in which dominant direction did object B (dark car with bright headlights on the right side of the track) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. backward\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "277c40c363e2fe3c13480cad554ed7826b8bb66cb443dfa0f6a28b0f94802587"}
{"qa_id": "youtube_vis__f3b5e6c84a__camera_object_distance_change__03", "video_id": "youtube_vis__f3b5e6c84a", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object B (dark car with bright headlights on the right side of the track) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/db9fda5637b803d752ba/F0.jpg", "media/db9fda5637b803d752ba/F1.jpg", "media/db9fda5637b803d752ba/F2.jpg", "media/db9fda5637b803d752ba/F3.jpg", "media/db9fda5637b803d752ba/F4.jpg", "media/db9fda5637b803d752ba/F5.jpg", "media/db9fda5637b803d752ba/F6.jpg", "media/db9fda5637b803d752ba/F7.jpg", "media/db9fda5637b803d752ba/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object B (dark car with bright headlights on the right side of the track) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "25ff7ca4c237a0225c5597dc985627b8727a5cdccce3d41e22345d0ec69147d1"}
{"qa_id": "youtube_vis__f4dd51ac35__object_world_motion_direction__00", "video_id": "youtube_vis__f4dd51ac35", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in black shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/f4c0d9fe332b0160cd6c/F0.jpg", "media/f4c0d9fe332b0160cd6c/F1.jpg", "media/f4c0d9fe332b0160cd6c/F2.jpg", "media/f4c0d9fe332b0160cd6c/F3.jpg", "media/f4c0d9fe332b0160cd6c/F4.jpg", "media/f4c0d9fe332b0160cd6c/F5.jpg", "media/f4c0d9fe332b0160cd6c/F6.jpg", "media/f4c0d9fe332b0160cd6c/F7.jpg", "media/f4c0d9fe332b0160cd6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person in black shirt riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9c7e4570a54eebf7288db06540aac51c4d745a220a9265ce36262347978b4583"}
{"qa_id": "youtube_vis__f4dd51ac35__camera_object_distance_change__01", "video_id": "youtube_vis__f4dd51ac35", "task": "camera_object_distance_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, did object A (person in black shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/f4c0d9fe332b0160cd6c/F0.jpg", "media/f4c0d9fe332b0160cd6c/F1.jpg", "media/f4c0d9fe332b0160cd6c/F2.jpg", "media/f4c0d9fe332b0160cd6c/F3.jpg", "media/f4c0d9fe332b0160cd6c/F4.jpg", "media/f4c0d9fe332b0160cd6c/F5.jpg", "media/f4c0d9fe332b0160cd6c/F6.jpg", "media/f4c0d9fe332b0160cd6c/F7.jpg", "media/f4c0d9fe332b0160cd6c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, did object A (person in black shirt riding a bicycle) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "40b5152b2d0e35240c64ee5513741701d259ac69452c6d528cfece0460a2171d"}
{"qa_id": "youtube_vis__f6adb12c42__camera_translation_direction__00", "video_id": "youtube_vis__f6adb12c42", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "down", "backward", "forward", "right", "up"], "frame_paths": ["media/0f7eb138b91b5cc2f951/F0.jpg", "media/0f7eb138b91b5cc2f951/F1.jpg", "media/0f7eb138b91b5cc2f951/F2.jpg", "media/0f7eb138b91b5cc2f951/F3.jpg", "media/0f7eb138b91b5cc2f951/F4.jpg", "media/0f7eb138b91b5cc2f951/F5.jpg", "media/0f7eb138b91b5cc2f951/F6.jpg", "media/0f7eb138b91b5cc2f951/F7.jpg", "media/0f7eb138b91b5cc2f951/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F3, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. down\nC. backward\nD. forward\nE. right\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "0938bb161f190102e82f5d97bd6c20908acc167f418c2d4017b5dfc08e1d234f"}
{"qa_id": "youtube_vis__f6adb12c42__object_world_motion_direction__01", "video_id": "youtube_vis__f6adb12c42", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F0 to F3, in which dominant direction did object A (person with long blonde hair wearing a yellow t-shirt and dark pants riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["backward", "right", "down", "up", "left", "forward"], "frame_paths": ["media/1b482f2b85263779b374/F0.jpg", "media/1b482f2b85263779b374/F1.jpg", "media/1b482f2b85263779b374/F2.jpg", "media/1b482f2b85263779b374/F3.jpg", "media/1b482f2b85263779b374/F4.jpg", "media/1b482f2b85263779b374/F5.jpg", "media/1b482f2b85263779b374/F6.jpg", "media/1b482f2b85263779b374/F7.jpg", "media/1b482f2b85263779b374/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F0 to F3, in which dominant direction did object A (person with long blonde hair wearing a yellow t-shirt and dark pants riding a bicycle) move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. backward\nB. right\nC. down\nD. up\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "060ac7d01d6b5f2000d966fe5666b62b1612ce763e0e97b3b3629c4de29ddea2"}
{"qa_id": "youtube_vis__f82b53017e__camera_object_distance_change__00", "video_id": "youtube_vis__f82b53017e", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F5 to F8, did object B (person in light blue jacket and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/efde89491678865cdc29/F0.jpg", "media/efde89491678865cdc29/F1.jpg", "media/efde89491678865cdc29/F2.jpg", "media/efde89491678865cdc29/F3.jpg", "media/efde89491678865cdc29/F4.jpg", "media/efde89491678865cdc29/F5.jpg", "media/efde89491678865cdc29/F6.jpg", "media/efde89491678865cdc29/F7.jpg", "media/efde89491678865cdc29/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F5 to F8, did object B (person in light blue jacket and orange pants) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "a83a495d4c9a913780d5e29de15b746b942cbb153d1b5b735947298b5dd54925"}
{"qa_id": "youtube_vis__f8d9c1eaf7__camera_translation_direction__00", "video_id": "youtube_vis__f8d9c1eaf7", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "down", "left"], "frame_paths": ["media/3d0eb954312c2cb3bf98/F0.jpg", "media/3d0eb954312c2cb3bf98/F1.jpg", "media/3d0eb954312c2cb3bf98/F2.jpg", "media/3d0eb954312c2cb3bf98/F3.jpg", "media/3d0eb954312c2cb3bf98/F4.jpg", "media/3d0eb954312c2cb3bf98/F5.jpg", "media/3d0eb954312c2cb3bf98/F6.jpg", "media/3d0eb954312c2cb3bf98/F7.jpg", "media/3d0eb954312c2cb3bf98/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "1ec3c1bf26c687b99ec7a839fcf55dafaae6c9c5288719bd5e415069104f25aa"}
{"qa_id": "youtube_vis__f8d9c1eaf7__object_world_motion_direction__01", "video_id": "youtube_vis__f8d9c1eaf7", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in black wetsuit wakeboarding) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.", "options": ["down", "left", "up", "forward", "backward", "right"], "frame_paths": ["media/9420e8e7954a085fcd51/F0.jpg", "media/9420e8e7954a085fcd51/F1.jpg", "media/9420e8e7954a085fcd51/F2.jpg", "media/9420e8e7954a085fcd51/F3.jpg", "media/9420e8e7954a085fcd51/F4.jpg", "media/9420e8e7954a085fcd51/F5.jpg", "media/9420e8e7954a085fcd51/F6.jpg", "media/9420e8e7954a085fcd51/F7.jpg", "media/9420e8e7954a085fcd51/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F3 to F8, in which dominant direction did object A (person in black wetsuit wakeboarding) move? Use the camera axes frozen at F3: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. up\nD. forward\nE. backward\nF. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "15d686f93fb315ea5e07e50f4ba1deb022eeaaf3dc0dda8f5aac5e0991208079"}
{"qa_id": "youtube_vis__f8d9c1eaf7__camera_object_distance_change__02", "video_id": "youtube_vis__f8d9c1eaf7", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F3 to F8, did object A (person in black wetsuit wakeboarding) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/9420e8e7954a085fcd51/F0.jpg", "media/9420e8e7954a085fcd51/F1.jpg", "media/9420e8e7954a085fcd51/F2.jpg", "media/9420e8e7954a085fcd51/F3.jpg", "media/9420e8e7954a085fcd51/F4.jpg", "media/9420e8e7954a085fcd51/F5.jpg", "media/9420e8e7954a085fcd51/F6.jpg", "media/9420e8e7954a085fcd51/F7.jpg", "media/9420e8e7954a085fcd51/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F3 to F8, did object A (person in black wetsuit wakeboarding) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "06880e0fe6655f7e96a33a668eebe0a47ab90bf607ec8305233719a34ee4355b"}
{"qa_id": "youtube_vis__f8fb020f47__camera_translation_direction__00", "video_id": "youtube_vis__f8fb020f47", "task": "camera_translation_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "left", "down"], "frame_paths": ["media/2f5437a27e20083f0484/F0.jpg", "media/2f5437a27e20083f0484/F1.jpg", "media/2f5437a27e20083f0484/F2.jpg", "media/2f5437a27e20083f0484/F3.jpg", "media/2f5437a27e20083f0484/F4.jpg", "media/2f5437a27e20083f0484/F5.jpg", "media/2f5437a27e20083f0484/F6.jpg", "media/2f5437a27e20083f0484/F7.jpg", "media/2f5437a27e20083f0484/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "143f5967be53ebe2f3332a8c848724300a7825ee38b61800cc89f2b8afefc089"}
{"qa_id": "youtube_vis__f8fb020f47__object_world_motion_direction__01", "video_id": "youtube_vis__f8fb020f47", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F4 to F6, in which dominant direction did object A (cyclist in black jacket and helmet) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.", "options": ["down", "left", "backward", "up", "right", "forward"], "frame_paths": ["media/646d4a32249656c0333c/F0.jpg", "media/646d4a32249656c0333c/F1.jpg", "media/646d4a32249656c0333c/F2.jpg", "media/646d4a32249656c0333c/F3.jpg", "media/646d4a32249656c0333c/F4.jpg", "media/646d4a32249656c0333c/F5.jpg", "media/646d4a32249656c0333c/F6.jpg", "media/646d4a32249656c0333c/F7.jpg", "media/646d4a32249656c0333c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F4 to F6, in which dominant direction did object A (cyclist in black jacket and helmet) move? Use the camera axes frozen at F4: right/down/forward are +x/+y/+z.\n\nOptions:\nA. down\nB. left\nC. backward\nD. up\nE. right\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62210a016880ac29310d79db4d203cb021d717672b947cead4f7989b2e436835"}
{"qa_id": "youtube_vis__f9cac1e19b__object_world_motion_direction__00", "video_id": "youtube_vis__f9cac1e19b", "task": "object_world_motion_direction", "group": "LLH", "group4": "LLH", "question": "After compensating for camera motion from F6 to F8, in which dominant direction did object A (white truck cab) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.", "options": ["right", "left", "up", "down", "backward", "forward"], "frame_paths": ["media/214d8f732bff6bf9a0d6/F0.jpg", "media/214d8f732bff6bf9a0d6/F1.jpg", "media/214d8f732bff6bf9a0d6/F2.jpg", "media/214d8f732bff6bf9a0d6/F3.jpg", "media/214d8f732bff6bf9a0d6/F4.jpg", "media/214d8f732bff6bf9a0d6/F5.jpg", "media/214d8f732bff6bf9a0d6/F6.jpg", "media/214d8f732bff6bf9a0d6/F7.jpg", "media/214d8f732bff6bf9a0d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F6 to F8, in which dominant direction did object A (white truck cab) move? Use the camera axes frozen at F6: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. up\nD. down\nE. backward\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "62aac73249d2d52be701a9e9b179509fe5b9f932130c0877e37088417eab94da"}
{"qa_id": "youtube_vis__f9cac1e19b__camera_object_distance_change__01", "video_id": "youtube_vis__f9cac1e19b", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F6 to F8, did object A (white truck cab) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/214d8f732bff6bf9a0d6/F0.jpg", "media/214d8f732bff6bf9a0d6/F1.jpg", "media/214d8f732bff6bf9a0d6/F2.jpg", "media/214d8f732bff6bf9a0d6/F3.jpg", "media/214d8f732bff6bf9a0d6/F4.jpg", "media/214d8f732bff6bf9a0d6/F5.jpg", "media/214d8f732bff6bf9a0d6/F6.jpg", "media/214d8f732bff6bf9a0d6/F7.jpg", "media/214d8f732bff6bf9a0d6/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g4"}, "requested_object_ids": ["g4"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F6 to F8, did object A (white truck cab) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "958c5d12dfc273a53db65d6c1b288fdb69d3511f5af80464cfa346320329d548"}
{"qa_id": "youtube_vis__f9f3852526__camera_view_direction_change__00", "video_id": "youtube_vis__f9f3852526", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/5a7438b30e35fc797120/F0.jpg", "media/5a7438b30e35fc797120/F1.jpg", "media/5a7438b30e35fc797120/F2.jpg", "media/5a7438b30e35fc797120/F3.jpg", "media/5a7438b30e35fc797120/F4.jpg", "media/5a7438b30e35fc797120/F5.jpg", "media/5a7438b30e35fc797120/F6.jpg", "media/5a7438b30e35fc797120/F7.jpg", "media/5a7438b30e35fc797120/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F4, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "dcea6da4b2f031518ed8c356ea0f756ba63782cf351f89f1f2d49381c78deedf"}
{"qa_id": "youtube_vis__fa04c615cf__object_world_motion_direction__00", "video_id": "youtube_vis__fa04c615cf", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark clothing riding a skateboard) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "right", "down", "left", "forward"], "frame_paths": ["media/e5a76ba66384ff17ce74/F0.jpg", "media/e5a76ba66384ff17ce74/F1.jpg", "media/e5a76ba66384ff17ce74/F2.jpg", "media/e5a76ba66384ff17ce74/F3.jpg", "media/e5a76ba66384ff17ce74/F4.jpg", "media/e5a76ba66384ff17ce74/F5.jpg", "media/e5a76ba66384ff17ce74/F6.jpg", "media/e5a76ba66384ff17ce74/F7.jpg", "media/e5a76ba66384ff17ce74/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in dark clothing riding a skateboard) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. right\nD. down\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "334e20802b1d1d947f8eae405c70f749a9f08a40cf0dce8c1f68cc4d78a7d955"}
{"qa_id": "youtube_vis__fbd77444cd__camera_translation_direction__00", "video_id": "youtube_vis__fbd77444cd", "task": "camera_translation_direction", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["left", "forward", "backward", "up", "right", "down"], "frame_paths": ["media/78a13621adc90199bf0a/F0.jpg", "media/78a13621adc90199bf0a/F1.jpg", "media/78a13621adc90199bf0a/F2.jpg", "media/78a13621adc90199bf0a/F3.jpg", "media/78a13621adc90199bf0a/F4.jpg", "media/78a13621adc90199bf0a/F5.jpg", "media/78a13621adc90199bf0a/F6.jpg", "media/78a13621adc90199bf0a/F7.jpg", "media/78a13621adc90199bf0a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. left\nB. forward\nC. backward\nD. up\nE. right\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "45935e8f5fab03ee57c2db890df3af845d90416db65e3d7b6bcce4aff3bd175e"}
{"qa_id": "youtube_vis__fbd77444cd__camera_object_distance_change__01", "video_id": "youtube_vis__fbd77444cd", "task": "camera_object_distance_change", "group": "HLL", "group4": "HLL+LHL+HHL", "question": "From F0 to F2, did object A (small grey dog with pointy ears) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/d9a5d89498d74ac4726a/F0.jpg", "media/d9a5d89498d74ac4726a/F1.jpg", "media/d9a5d89498d74ac4726a/F2.jpg", "media/d9a5d89498d74ac4726a/F3.jpg", "media/d9a5d89498d74ac4726a/F4.jpg", "media/d9a5d89498d74ac4726a/F5.jpg", "media/d9a5d89498d74ac4726a/F6.jpg", "media/d9a5d89498d74ac4726a/F7.jpg", "media/d9a5d89498d74ac4726a/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F2, did object A (small grey dog with pointy ears) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "725b55172d483067053dedb4f9a0e7810b68a65be157b53ba70c75aaf12f8e78"}
{"qa_id": "youtube_vis__fbe53dc8e8__camera_view_direction_change__00", "video_id": "youtube_vis__fbe53dc8e8", "task": "camera_view_direction_change", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?", "options": ["right", "up", "down", "left"], "frame_paths": ["media/acce10b8565035a369dd/F0.jpg", "media/acce10b8565035a369dd/F1.jpg", "media/acce10b8565035a369dd/F2.jpg", "media/acce10b8565035a369dd/F3.jpg", "media/acce10b8565035a369dd/F4.jpg", "media/acce10b8565035a369dd/F5.jpg", "media/acce10b8565035a369dd/F6.jpg", "media/acce10b8565035a369dd/F7.jpg", "media/acce10b8565035a369dd/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F2?\n\nOptions:\nA. right\nB. up\nC. down\nD. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6eae66ee8c9fd961541d1dbc1fbfab09c9704c7eed5843b6a680b6012297addc"}
{"qa_id": "youtube_vis__fbe53dc8e8__object_world_motion_direction__01", "video_id": "youtube_vis__fbe53dc8e8", "task": "object_world_motion_direction", "group": "HHH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in blue jeans riding a skateboard) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["right", "backward", "up", "forward", "down", "left"], "frame_paths": ["media/30117f2dcc1d2e5aa057/F0.jpg", "media/30117f2dcc1d2e5aa057/F1.jpg", "media/30117f2dcc1d2e5aa057/F2.jpg", "media/30117f2dcc1d2e5aa057/F3.jpg", "media/30117f2dcc1d2e5aa057/F4.jpg", "media/30117f2dcc1d2e5aa057/F5.jpg", "media/30117f2dcc1d2e5aa057/F6.jpg", "media/30117f2dcc1d2e5aa057/F7.jpg", "media/30117f2dcc1d2e5aa057/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (person in blue jeans riding a skateboard) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. backward\nC. up\nD. forward\nE. down\nF. left\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "4185cefad7958e3d0afd85b6eecced1cf3f5efac2c2e5a0cf3429ad47c1ffd90"}
{"qa_id": "youtube_vis__fd3dd6202e__object_world_motion_direction__00", "video_id": "youtube_vis__fd3dd6202e", "task": "object_world_motion_direction", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "After compensating for camera motion from F2 to F5, in which dominant direction did object A (blue car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.", "options": ["up", "backward", "forward", "right", "left", "down"], "frame_paths": ["media/bdd676d2a2dfa7ecbed9/F0.jpg", "media/bdd676d2a2dfa7ecbed9/F1.jpg", "media/bdd676d2a2dfa7ecbed9/F2.jpg", "media/bdd676d2a2dfa7ecbed9/F3.jpg", "media/bdd676d2a2dfa7ecbed9/F4.jpg", "media/bdd676d2a2dfa7ecbed9/F5.jpg", "media/bdd676d2a2dfa7ecbed9/F6.jpg", "media/bdd676d2a2dfa7ecbed9/F7.jpg", "media/bdd676d2a2dfa7ecbed9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F2 to F5, in which dominant direction did object A (blue car) move? Use the camera axes frozen at F2: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. backward\nC. forward\nD. right\nE. left\nF. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "b33c90014d6145e17a777c0ecae25c6104b806c4c271fa10d6278849c4841840"}
{"qa_id": "youtube_vis__fd3dd6202e__camera_object_distance_change__01", "video_id": "youtube_vis__fd3dd6202e", "task": "camera_object_distance_change", "group": "HLH", "group4": "HLH+LHH+HHH", "question": "From F2 to F5, did object A (blue car) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/bdd676d2a2dfa7ecbed9/F0.jpg", "media/bdd676d2a2dfa7ecbed9/F1.jpg", "media/bdd676d2a2dfa7ecbed9/F2.jpg", "media/bdd676d2a2dfa7ecbed9/F3.jpg", "media/bdd676d2a2dfa7ecbed9/F4.jpg", "media/bdd676d2a2dfa7ecbed9/F5.jpg", "media/bdd676d2a2dfa7ecbed9/F6.jpg", "media/bdd676d2a2dfa7ecbed9/F7.jpg", "media/bdd676d2a2dfa7ecbed9/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F2 to F5, did object A (blue car) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "332a4e957fb7fbf9c4288a39dd972c15878743c8787415d3528d0cdcef4cd702"}
{"qa_id": "youtube_vis__fdb28d0fbb__camera_translation_direction__00", "video_id": "youtube_vis__fdb28d0fbb", "task": "camera_translation_direction", "group": "LLL", "group4": "LLL", "question": "From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.", "options": ["right", "left", "down", "forward", "backward", "up"], "frame_paths": ["media/ffb97bacf452db08f348/F0.jpg", "media/ffb97bacf452db08f348/F1.jpg", "media/ffb97bacf452db08f348/F2.jpg", "media/ffb97bacf452db08f348/F3.jpg", "media/ffb97bacf452db08f348/F4.jpg", "media/ffb97bacf452db08f348/F5.jpg", "media/ffb97bacf452db08f348/F6.jpg", "media/ffb97bacf452db08f348/F7.jpg", "media/ffb97bacf452db08f348/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F8, in which dominant direction did the camera center move? Use the camera axes frozen at F0: right/down/forward are +x/+y/+z.\n\nOptions:\nA. right\nB. left\nC. down\nD. forward\nE. backward\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "65604af0fa63a051666412f62355095e2592d8e3cdae8d12aefd16c7116cb6c0"}
{"qa_id": "youtube_vis__fdb28d0fbb__object_world_motion_direction__01", "video_id": "youtube_vis__fdb28d0fbb", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in dark clothing standing on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["up", "down", "right", "backward", "left", "forward"], "frame_paths": ["media/9c8361e12a89d9848330/F0.jpg", "media/9c8361e12a89d9848330/F1.jpg", "media/9c8361e12a89d9848330/F2.jpg", "media/9c8361e12a89d9848330/F3.jpg", "media/9c8361e12a89d9848330/F4.jpg", "media/9c8361e12a89d9848330/F5.jpg", "media/9c8361e12a89d9848330/F6.jpg", "media/9c8361e12a89d9848330/F7.jpg", "media/9c8361e12a89d9848330/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object A (person in dark clothing standing on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. up\nB. down\nC. right\nD. backward\nE. left\nF. forward\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "04bf1a6b7ca1895a15defeb01b50b7c8e21b5d495252c1d360a5d9d9a956dc36"}
{"qa_id": "youtube_vis__fdb28d0fbb__object_world_motion_direction__02", "video_id": "youtube_vis__fdb28d0fbb", "task": "object_world_motion_direction", "group": "LLL", "group4": "LLL", "question": "After compensating for camera motion from F1 to F8, in which dominant direction did object B (person in dark clothing standing on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.", "options": ["forward", "right", "backward", "left", "down", "up"], "frame_paths": ["media/fbd156e71f35083ee878/F0.jpg", "media/fbd156e71f35083ee878/F1.jpg", "media/fbd156e71f35083ee878/F2.jpg", "media/fbd156e71f35083ee878/F3.jpg", "media/fbd156e71f35083ee878/F4.jpg", "media/fbd156e71f35083ee878/F5.jpg", "media/fbd156e71f35083ee878/F6.jpg", "media/fbd156e71f35083ee878/F7.jpg", "media/fbd156e71f35083ee878/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: After compensating for camera motion from F1 to F8, in which dominant direction did object B (person in dark clothing standing on snow) move? Use the camera axes frozen at F1: right/down/forward are +x/+y/+z.\n\nOptions:\nA. forward\nB. right\nC. backward\nD. left\nE. down\nF. up\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "087a968f7587011749c475d4531e21423ab99de70617f643ade5498437cbbe73"}
{"qa_id": "youtube_vis__fdc6e3d581__camera_view_direction_change__00", "video_id": "youtube_vis__fdc6e3d581", "task": "camera_view_direction_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["up", "right", "left", "down"], "frame_paths": ["media/5bb2df58c108580bfaa2/F0.jpg", "media/5bb2df58c108580bfaa2/F1.jpg", "media/5bb2df58c108580bfaa2/F2.jpg", "media/5bb2df58c108580bfaa2/F3.jpg", "media/5bb2df58c108580bfaa2/F4.jpg", "media/5bb2df58c108580bfaa2/F5.jpg", "media/5bb2df58c108580bfaa2/F6.jpg", "media/5bb2df58c108580bfaa2/F7.jpg", "media/5bb2df58c108580bfaa2/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. up\nB. right\nC. left\nD. down\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "c5d79ec53522c454e1712763d5353853f0417ac1d325bbb29245cb0f985fe955"}
{"qa_id": "youtube_vis__fdc6e3d581__camera_object_distance_change__01", "video_id": "youtube_vis__fdc6e3d581", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object A (red two-wheeled farm implement) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/450dabf8f1e91e04be1c/F0.jpg", "media/450dabf8f1e91e04be1c/F1.jpg", "media/450dabf8f1e91e04be1c/F2.jpg", "media/450dabf8f1e91e04be1c/F3.jpg", "media/450dabf8f1e91e04be1c/F4.jpg", "media/450dabf8f1e91e04be1c/F5.jpg", "media/450dabf8f1e91e04be1c/F6.jpg", "media/450dabf8f1e91e04be1c/F7.jpg", "media/450dabf8f1e91e04be1c/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object A (red two-wheeled farm implement) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "398ef1d914ab6e691945e7386223cc180fd9988cbd0714621cef72d399198ac3"}
{"qa_id": "youtube_vis__fdc6e3d581__camera_object_distance_change__02", "video_id": "youtube_vis__fdc6e3d581", "task": "camera_object_distance_change", "group": "LHH", "group4": "HLH+LHH+HHH", "question": "From F0 to F5, did object B (person in white shirt and green hat) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["farther", "closer"], "frame_paths": ["media/c23a9746f133484e4ccf/F0.jpg", "media/c23a9746f133484e4ccf/F1.jpg", "media/c23a9746f133484e4ccf/F2.jpg", "media/c23a9746f133484e4ccf/F3.jpg", "media/c23a9746f133484e4ccf/F4.jpg", "media/c23a9746f133484e4ccf/F5.jpg", "media/c23a9746f133484e4ccf/F6.jpg", "media/c23a9746f133484e4ccf/F7.jpg", "media/c23a9746f133484e4ccf/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g2", "B": "g3"}, "requested_object_ids": ["g3"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F5, did object B (person in white shirt and green hat) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. farther\nB. closer\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6c8d4fd2dfeeae90e0bcdc58b14d0987678780a2019c978eed1ccbdc80d2364d"}
{"qa_id": "youtube_vis__fec5232a13__camera_view_direction_change__00", "video_id": "youtube_vis__fec5232a13", "task": "camera_view_direction_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?", "options": ["left", "up", "down", "right"], "frame_paths": ["media/935fd370e9dcead2000b/F0.jpg", "media/935fd370e9dcead2000b/F1.jpg", "media/935fd370e9dcead2000b/F2.jpg", "media/935fd370e9dcead2000b/F3.jpg", "media/935fd370e9dcead2000b/F4.jpg", "media/935fd370e9dcead2000b/F5.jpg", "media/935fd370e9dcead2000b/F6.jpg", "media/935fd370e9dcead2000b/F7.jpg", "media/935fd370e9dcead2000b/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": [], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, toward which direction did the camera optical axis turn, using the camera axes frozen at F0?\n\nOptions:\nA. left\nB. up\nC. down\nD. right\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "af75473fbfd63923e7a7dd9d31a1213d36f19af8df543626bae3d6af8373e9d0"}
{"qa_id": "youtube_vis__fec5232a13__camera_object_distance_change__01", "video_id": "youtube_vis__fec5232a13", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object A (black horse jumping over obstacles) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/e7ae43cc767788286715/F0.jpg", "media/e7ae43cc767788286715/F1.jpg", "media/e7ae43cc767788286715/F2.jpg", "media/e7ae43cc767788286715/F3.jpg", "media/e7ae43cc767788286715/F4.jpg", "media/e7ae43cc767788286715/F5.jpg", "media/e7ae43cc767788286715/F6.jpg", "media/e7ae43cc767788286715/F7.jpg", "media/e7ae43cc767788286715/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g1"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object A (black horse jumping over obstacles) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "9f881066a93e7c42f6ba763655ee8d549b067a90063365c4517ded673958a92f"}
{"qa_id": "youtube_vis__fec5232a13__camera_object_distance_change__02", "video_id": "youtube_vis__fec5232a13", "task": "camera_object_distance_change", "group": "LLH", "group4": "LLH", "question": "From F0 to F6, did object B (equestrian rider in white helmet and vest) become closer to or farther from the camera? Compare 3D Euclidean range.", "options": ["closer", "farther"], "frame_paths": ["media/6b1f006c13cd5eb0b6f7/F0.jpg", "media/6b1f006c13cd5eb0b6f7/F1.jpg", "media/6b1f006c13cd5eb0b6f7/F2.jpg", "media/6b1f006c13cd5eb0b6f7/F3.jpg", "media/6b1f006c13cd5eb0b6f7/F4.jpg", "media/6b1f006c13cd5eb0b6f7/F5.jpg", "media/6b1f006c13cd5eb0b6f7/F6.jpg", "media/6b1f006c13cd5eb0b6f7/F7.jpg", "media/6b1f006c13cd5eb0b6f7/F8.jpg"], "frame_indices": [0, 1, 2, 3, 4, 5, 6, 7, 8], "target_mapping": {"A": "g1", "B": "g2"}, "requested_object_ids": ["g2"], "prompt": "The nine images are ordered frames F0 through F8 of one released video sequence. Frame indices indicate order, not physical seconds. Orange contours mark target object A; green contours mark target object B, when those targets are requested. Use the frame range and coordinate convention in the question.\n\nQuestion: From F0 to F6, did object B (equestrian rider in white helmet and vest) become closer to or farther from the camera? Compare 3D Euclidean range.\n\nOptions:\nA. closer\nB. farther\n\nAnswer directly with exactly one option LETTER (A, B, C, D, E or F as applicable). The option letter is its position in the options list, not an object ID. Do not explain.", "prompt_sha256": "6f5602fd50806d31eccb00128557102f06d5f86f8eff9c9979f81cd501a2f0e7"}