Datasets:
type stringclasses 19
values | type_name stringclasses 19
values | dimension stringclasses 3
values | scene_id stringclasses 167
values | trajectory_name stringclasses 5
values | video stringclasses 835
values | delta_d_group stringclasses 5
values | question stringlengths 86 575 | correct_answer stringclasses 4
values | options dict | id stringlengths 27 36 | gt_value unknown | gt_secondary unknown | boundary_case bool 1
class | object_label stringlengths 2 54 ⌀ | validation_passed bool 1
class | task_type stringclasses 19
values | question_text stringlengths 66 353 | answer_text stringclasses 346
values | validation_status stringclasses 1
value | task_truth unknown | reward_targets dict | vsi_alignment dict | metadata_structured dict | metadata_structured_json stringlengths 359 695 | eval_prompt stringlengths 171 660 | paper_task_type stringclasses 19
values | training_task_type stringclasses 19
values | train_use_forbidden bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
B1 | displacement_range | kinematic_perception | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | C | {
"A": "Small (0.5 – 2 m)",
"B": "Large (> 5 m)",
"C": "Medium (2 – 5 m)",
"D": "Static (< 0.5 m)"
} | bench_00dd871005_fast_B1_000 | 3.0307 | null | false | null | true | displacement_range | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | Medium (2 – 5 m) | geometry_recomputed | {
"displacement_distance": 3.0307
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_distance": 3.0307
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_k... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_distance": 3.0307
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displace... | {"task_metadata":{"displacement_distance":3.0307},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"displacement_range","reward_kind":"kinematic_scalar","required_truth_fields":["displacement_distance"],"truth_fields":{"displacement_distance":3.0307},"apply_answer_reward":true,"apply_format_re... | <image>
These are frames of a video.
Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
Options:
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (<... | displacement_range | displacement_range | true |
B2 | camera_motion_direction | kinematic_perception | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear-Right
B. Rear-Left
C. Rear
D. Right
A. Front
B. Right
C. Front-Right
D. Left | D | {
"A": "Front",
"B": "Right",
"C": "Front-Right",
"D": "Left"
} | bench_00dd871005_fast_B2_000 | "Left" | {
"signed_theta_deg": -91.6394929893
} | false | null | true | camera_motion_direction | In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear-Right
B. Rear-Left
C. Rear
D. Right | Left | geometry_recomputed | {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "camera_motion_direction",
"reward_kind": "kinematic_direction",
"required_truth_fields": [
"displacement_vector",
"direction"
],
"truth_fields": {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"directi... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "camera_motion_direction",
"reward_kind": "kinematic_direction",
"required_truth_fields": [
... | {"task_metadata":{"displacement_vector":[-3.0291,-0.047,0.0867],"direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"camera_motion_direction","reward_kind":"kinematic_direction","required_truth_fields":["displacement_vector","direction"],"truth_fields":{"displacement_vector":[... | <image>
These are frames of a video.
In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear-Right
B. Rear-Left
C. Rear
D. Right
Options:
A. Front
B. Right
C. Front-Right
D. Left
Please answer with the option letter. | camera_motion_direction | camera_motion_direction | true |
B3 | rotation_range | kinematic_perception | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Large (> 90°)
B. Minimal (< 15°)
C. Medium (45 – 90°)
D. Small (15 – 45°)
A. Large (> 90°)
B. Minimal (< 15°)
C. Medium (45 – 90°)
D. Small (15 – 45°) | A | {
"A": "Large (> 90°)",
"B": "Minimal (< 15°)",
"C": "Medium (45 – 90°)",
"D": "Small (15 – 45°)"
} | bench_00dd871005_fast_B3_000 | 669.07 | null | false | null | true | rotation_range | What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Large (> 90°)
B. Minimal (< 15°)
C. Medium (45 – 90°)
D. Small (15 – 45°) | Large (> 90°) | geometry_recomputed | {
"rotation_angle": 669.07
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"rotation_angle"
],
"truth_fields": {
"rotation_angle": 669.07
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": ... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"rotation_angle": 669.07
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"rotation_angle"
],
"truth_fields": {
"rotation_angle": 669.07
... | {"task_metadata":{"rotation_angle":669.07},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"rotation_range","reward_kind":"kinematic_scalar","required_truth_fields":["rotation_angle"],"truth_fields":{"rotation_angle":669.07},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinemat... | <image>
These are frames of a video.
What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Large (> 90°)
B. Minimal (< 15°)
C. Medium (45 – 90°)
D. Small (15 – 45°)
Options:
A. Large (> 90°)
B. Minimal (< 15°)
C. Medium (45 – 90°)
D. Small (15 – 45°)
Please a... | rotation_range | rotation_range | true |
B6 | average_speed | kinematic_perception | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | What is the approximate average movement speed of the camera throughout this video?
A. Slow (0.5 – 1.5 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Moderate (1.5 – 3.0 m/s)
D. Fast (> 3.0 m/s)
A. Slow (0.5 – 1.5 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Moderate (1.5 – 3.0 m/s)
D. Fast (> 3.0 m/s) | A | {
"A": "Slow (0.5 – 1.5 m/s)",
"B": "Nearly stationary (< 0.5 m/s)",
"C": "Moderate (1.5 – 3.0 m/s)",
"D": "Fast (> 3.0 m/s)"
} | bench_00dd871005_fast_B6_000 | 1.29046 | null | false | null | true | average_speed | What is the approximate average movement speed of the camera throughout this video?
A. Slow (0.5 – 1.5 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Moderate (1.5 – 3.0 m/s)
D. Fast (> 3.0 m/s) | Slow (0.5 – 1.5 m/s) | geometry_recomputed | {
"average_speed": 1.29046
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "average_speed",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"average_speed"
],
"truth_fields": {
"average_speed": 1.29046
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": tr... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"average_speed": 1.29046
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "average_speed",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"average_speed"
],
"truth_fields": {
"average_speed": 1.29046
... | {"task_metadata":{"average_speed":1.29046},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"average_speed","reward_kind":"kinematic_scalar","required_truth_fields":["average_speed"],"truth_fields":{"average_speed":1.29046},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic... | <image>
These are frames of a video.
What is the approximate average movement speed of the camera throughout this video?
A. Slow (0.5 – 1.5 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Moderate (1.5 – 3.0 m/s)
D. Fast (> 3.0 m/s)
Options:
A. Slow (0.5 – 1.5 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Moderate (1.5 – 3.0 m/s)
... | average_speed | average_speed | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Front-Left
B. Rear-Left
C. Front-Right
D. Right
A. Front-Left
B. Rear-Left
C. Front-Right
D. Right | A | {
"A": "Front-Left",
"B": "Rear-Left",
"C": "Front-Right",
"D": "Right"
} | bench_00dd871005_fast_B7_001 | "Front-Left" | {
"distance_m": 1.0949746614,
"signed_theta_deg": -49.033259491
} | false | camping bag | true | first_frame_object_direction | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Front-Left
B. Rear-Left
C. Front-Right
D. Right | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"o... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_f... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Front-Left
B. Rear-Left
C. Front-Right
D. Right
Options:
A. Front-Left
B. Rear-Left
C. Front-Right
D. Right
Please answer with the option lette... | first_frame_object_direction | first_frame_object_direction | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front
C. Front-Left
D. Left
A. Rear-Left
B. Front
C. Front-Left
D. Left | B | {
"A": "Rear-Left",
"B": "Front",
"C": "Front-Left",
"D": "Left"
} | bench_00dd871005_fast_B7_003 | "Front" | {
"distance_m": 1.7479982957,
"signed_theta_deg": 7.3264066602
} | false | clothes | true | first_frame_object_direction | At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front
C. Front-Left
D. Left | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewa... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front
C. Front-Left
D. Left
Options:
A. Rear-Left
B. Front
C. Front-Left
D. Left
Please answer with the option letter. | first_frame_object_direction | first_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | At the end of the video, in which direction is the sofa relative to the camera?
A. Rear
B. Left
C. Front-Left
D. Front-Right
A. Rear
B. Left
C. Front-Left
D. Front-Right | B | {
"A": "Rear",
"B": "Left",
"C": "Front-Left",
"D": "Front-Right"
} | bench_00dd871005_fast_B8_001 | "Left" | {
"distance_m": 2.6230296135,
"signed_theta_deg": -80.2708269168
} | false | sofa | true | last_frame_object_direction | At the end of the video, in which direction is the sofa relative to the camera?
A. Rear
B. Left
C. Front-Left
D. Front-Right | Left | geometry_recomputed | {
"object_direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_k... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_d... | {"task_metadata":{"object_direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Left"},"apply_answer_reward":true,"apply_format_reward"... | <image>
These are frames of a video.
At the end of the video, in which direction is the sofa relative to the camera?
A. Rear
B. Left
C. Front-Left
D. Front-Right
Options:
A. Rear
B. Left
C. Front-Left
D. Front-Right
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Right
C. Rear-Right
D. Rear
A. Front-Left
B. Right
C. Rear-Right
D. Rear | A | {
"A": "Front-Left",
"B": "Right",
"C": "Rear-Right",
"D": "Rear"
} | bench_00dd871005_fast_B8_003 | "Front-Left" | {
"distance_m": 2.5161181989,
"signed_theta_deg": -40.7255937873
} | false | power sockets | true | last_frame_object_direction | At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Right
C. Rear-Right
D. Rear | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"a... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"ob... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_fo... | <image>
These are frames of a video.
At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Right
C. Rear-Right
D. Rear
Options:
A. Front-Left
B. Right
C. Rear-Right
D. Rear
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | At the end of the video, in which direction is the phone charger relative to the camera?
A. Rear-Right
B. Front-Right
C. Rear
D. Front-Left
A. Rear-Right
B. Front-Right
C. Rear
D. Front-Left | D | {
"A": "Rear-Right",
"B": "Front-Right",
"C": "Rear",
"D": "Front-Left"
} | bench_00dd871005_fast_B8_004 | "Front-Left" | {
"distance_m": 2.4899911107,
"signed_theta_deg": -41.5336090887
} | false | phone charger | true | last_frame_object_direction | At the end of the video, in which direction is the phone charger relative to the camera?
A. Rear-Right
B. Front-Right
C. Rear
D. Front-Left | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"a... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"ob... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_fo... | <image>
These are frames of a video.
At the end of the video, in which direction is the phone charger relative to the camera?
A. Rear-Right
B. Front-Right
C. Rear
D. Front-Left
Options:
A. Rear-Right
B. Front-Right
C. Rear
D. Front-Left
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. back-right
B. front-right
C. back-left
D. front-left | B | {
"A": "back-right",
"B": "front-right",
"C": "back-left",
"D": "front-left"
} | bench_00dd871005_fast_B11_000 | "front-right" | {
"proj_right": 1.1205207594,
"proj_forward": 0.6019209897000001,
"signed_theta_deg": 61.7562112133
} | false | bed,sofa,phone charger | true | coordinate_transform_direction | If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-right | geometry_recomputed | {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":61.75621121325139,"relative_direction":"front-right"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sign... | <image>
These are frames of a video.
If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. back-right
B. front-right
C... | coordinate_transform_direction | coordinate_transform_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. front-left
B. back-left
C. front-right
D. back-right | C | {
"A": "front-left",
"B": "back-left",
"C": "front-right",
"D": "back-right"
} | bench_00dd871005_fast_B11_001 | "front-right" | {
"proj_right": 0.9021692509,
"proj_forward": 0.9590482901,
"signed_theta_deg": 43.2495786434
} | false | bed,sofa,projector | true | coordinate_transform_direction | If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-right | geometry_recomputed | {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":43.249578643434816,"relative_direction":"front-right"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sig... | <image>
These are frames of a video.
If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. front-left
B. back-left
C. fron... | coordinate_transform_direction | coordinate_transform_direction | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | Measuring from the closest point of each object, what is the approximate distance between fruit and sponge?
A. Very close (< 1 m)
B. Far (> 6 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m) | D | {
"A": "Very close (< 1 m)",
"B": "Far (> 6 m)",
"C": "Medium (3 – 6 m)",
"D": "Close (1 – 3 m)"
} | bench_00dd871005_fast_B12_002 | 1.1780487340999999 | null | false | fruit,sponge | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between fruit and sponge? | Close (1 – 3 m) | geometry_recomputed | {
"aabb_surface_distance_m": 1.1780487340999999
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 1.1780487340999999
},
"apply_answer_reward": true,
"apply_format_reward... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 1.1780487340999999
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields"... | {"task_metadata":{"aabb_surface_distance_m":1.1780487341362413},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":1.1780487341362413},"apply_ans... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between fruit and sponge?
Options:
A. Very close (< 1 m)
B. Far (> 6 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B13 | object_count | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | How many spice rack objects are visible in this scene?
A. 3
B. 4
C. 1
D. 2
A. 3
B. 4
C. 1
D. 2 | C | {
"A": "3",
"B": "4",
"C": "1",
"D": "2"
} | bench_00dd871005_fast_B13_000 | 1 | null | false | spice rack | true | object_count | How many spice rack objects are visible in this scene?
A. 3
B. 4
C. 1
D. 2 | 1 | geometry_recomputed | {
"object_count": 1
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": false,
"prov... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_count": 1
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answe... | {"task_metadata":{"object_count":1},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_count","reward_kind":"spatial_count","required_truth_fields":["object_count"],"truth_fields":{"object_count":1},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic_reward":false,"pro... | <image>
These are frames of a video.
How many spice rack objects are visible in this scene?
A. 3
B. 4
C. 1
D. 2
Options:
A. 3
B. 4
C. 1
D. 2
Please answer with the option letter. | object_count | object_count | true |
B13 | object_count | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | How many kitchen counter objects are visible in this scene?
A. 1
B. 3
C. 4
D. 2
A. 1
B. 3
C. 4
D. 2 | A | {
"A": "1",
"B": "3",
"C": "4",
"D": "2"
} | bench_00dd871005_fast_B13_002 | 1 | null | false | kitchen counter | true | object_count | How many kitchen counter objects are visible in this scene?
A. 1
B. 3
C. 4
D. 2 | 1 | geometry_recomputed | {
"object_count": 1
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": false,
"prov... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_count": 1
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answe... | {"task_metadata":{"object_count":1},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_count","reward_kind":"spatial_count","required_truth_fields":["object_count"],"truth_fields":{"object_count":1},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic_reward":false,"pro... | <image>
These are frames of a video.
How many kitchen counter objects are visible in this scene?
A. 1
B. 3
C. 4
D. 2
Options:
A. 1
B. 3
C. 4
D. 2
Please answer with the option letter. | object_count | object_count | true |
B14 | object_size | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | What is the longest dimension of the suitcase's bounding box?
A. Small (< 0.3 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Large (1.0 – 2.0 m)
A. Small (< 0.3 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Large (1.0 – 2.0 m) | C | {
"A": "Small (< 0.3 m)",
"B": "Very large (> 2.0 m)",
"C": "Medium (0.3 – 1.0 m)",
"D": "Large (1.0 – 2.0 m)"
} | bench_00dd871005_fast_B14_000 | 0.3995 | [
0.2187,
0.3995,
0.3687
] | false | suitcase | true | object_size | What is the longest dimension of the suitcase's bounding box?
A. Small (< 0.3 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Large (1.0 – 2.0 m) | Medium (0.3 – 1.0 m) | geometry_recomputed | {
"aabb_extents_m": [
0.2187,
0.3995,
0.3687
],
"longest_extent_m": 0.3995
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.2187,
0.3995,
0.3687
],
"longest_extent_m": 0.3995
},... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.2187,
0.3995,
0.3687
],
"longest_extent_m": 0.3995
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extent... | {"task_metadata":{"aabb_extents_m":[0.21870000000000012,0.3995000000000002,0.3687],"longest_extent_m":0.3995000000000002},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth_fields":{"a... | <image>
These are frames of a video.
What is the longest dimension of the suitcase's bounding box?
A. Small (< 0.3 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Large (1.0 – 2.0 m)
Options:
A. Small (< 0.3 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Large (1.0 – 2.0 m)
Please answer with the option l... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | What is the longest dimension of the water pitcher's bounding box?
A. Large (1.0 – 2.0 m)
B. Small (< 0.3 m)
C. Very large (> 2.0 m)
D. Medium (0.3 – 1.0 m)
A. Large (1.0 – 2.0 m)
B. Small (< 0.3 m)
C. Very large (> 2.0 m)
D. Medium (0.3 – 1.0 m) | B | {
"A": "Large (1.0 – 2.0 m)",
"B": "Small (< 0.3 m)",
"C": "Very large (> 2.0 m)",
"D": "Medium (0.3 – 1.0 m)"
} | bench_00dd871005_fast_B14_001 | 0.21380000000000002 | [
0.1743,
0.1391,
0.21380000000000002
] | false | water pitcher | true | object_size | What is the longest dimension of the water pitcher's bounding box?
A. Large (1.0 – 2.0 m)
B. Small (< 0.3 m)
C. Very large (> 2.0 m)
D. Medium (0.3 – 1.0 m) | Small (< 0.3 m) | geometry_recomputed | {
"aabb_extents_m": [
0.1743,
0.1391,
0.21380000000000002
],
"longest_extent_m": 0.21380000000000002
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.1743,
0.1391,
0.21380000000000002
],
"longest_extent_m"... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.1743,
0.1391,
0.21380000000000002
],
"longest_extent_m": 0.21380000000000002
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fie... | {"task_metadata":{"aabb_extents_m":[0.17429999999999968,0.1391,0.2138000000000001],"longest_extent_m":0.2138000000000001},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth_fields":{"a... | <image>
These are frames of a video.
What is the longest dimension of the water pitcher's bounding box?
A. Large (1.0 – 2.0 m)
B. Small (< 0.3 m)
C. Very large (> 2.0 m)
D. Medium (0.3 – 1.0 m)
Options:
A. Large (1.0 – 2.0 m)
B. Small (< 0.3 m)
C. Very large (> 2.0 m)
D. Medium (0.3 – 1.0 m)
Please answer with the opt... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | What is the longest dimension of the clothes basket's bounding box?
A. Medium (0.3 – 1.0 m)
B. Large (1.0 – 2.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
A. Medium (0.3 – 1.0 m)
B. Large (1.0 – 2.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | C | {
"A": "Medium (0.3 – 1.0 m)",
"B": "Large (1.0 – 2.0 m)",
"C": "Small (< 0.3 m)",
"D": "Very large (> 2.0 m)"
} | bench_00dd871005_fast_B14_002 | 0.2691 | [
0.2169,
0.1913,
0.2691
] | false | clothes basket | true | object_size | What is the longest dimension of the clothes basket's bounding box?
A. Medium (0.3 – 1.0 m)
B. Large (1.0 – 2.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | Small (< 0.3 m) | geometry_recomputed | {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
},... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extent... | {"task_metadata":{"aabb_extents_m":[0.21689999999999987,0.19130000000000003,0.2691],"longest_extent_m":0.2691},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth_fields":{"aabb_extents... | <image>
These are frames of a video.
What is the longest dimension of the clothes basket's bounding box?
A. Medium (0.3 – 1.0 m)
B. Large (1.0 – 2.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Options:
A. Medium (0.3 – 1.0 m)
B. Large (1.0 – 2.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Please answer with the op... | object_size | object_size | true |
B17 | motion_type | implicit_trajectory_reasoning | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Stationary rotation (looking around without moving)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction)
A. Static (no significant movement)... | C | {
"A": "Static (no significant movement)",
"B": "Stationary rotation (looking around without moving)",
"C": "Curved path with both translation and rotation",
"D": "Straight-line translation (moving in one direction)"
} | bench_00dd871005_fast_B17_000 | {
"straightness": 0.46970847600000004,
"rotation_deg": 669.07,
"disp": 3.0307
} | null | false | null | true | motion_type | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Stationary rotation (looking around without moving)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction) | Curved path with both translation and rotation | geometry_recomputed | {
"straightness": 0.46970847600000004,
"rotation_deg": 669.07,
"disp": 3.0307
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"rotation_deg",
"disp"
],
"truth_fields": {
"straightness": 0.46970847600000004,
"rotation_deg": 669.07,
"disp": 3.0307
},
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"straightness": 0.46970847600000004,
"rotation_deg": 669.07,
"disp": 3.0307
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
... | {"task_metadata":{"straightness":0.4697084760473009,"rotation_deg":669.07,"disp":3.0307},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"motion_type","reward_kind":"kinematic_composite","required_truth_fields":["straightness","rotation_deg","disp"],"truth_fields":{"straightness":0.4697084760... | <image>
These are frames of a video.
Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Stationary rotation (looking around without moving)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction... | motion_type | motion_type | true |
B18 | trajectory_shape | implicit_trajectory_reasoning | 00dd871005 | fast | 00dd871005_hq/video_fast.mp4 | 2.0m | Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. L-shaped path (one major turn)
C. U-shaped / returning path
D. Approximately straight line
A. Curved / irregular path
B. L-shaped path (one major turn)
C. U-shaped / return... | A | {
"A": "Curved / irregular path",
"B": "L-shaped path (one major turn)",
"C": "U-shaped / returning path",
"D": "Approximately straight line"
} | bench_00dd871005_fast_B18_000 | {
"straightness": 0.46970847600000004,
"max_turn_angle": 33.5774641649,
"closure_ratio": 0.9999653752000001
} | null | false | null | true | trajectory_shape | Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. L-shaped path (one major turn)
C. U-shaped / returning path
D. Approximately straight line | Curved / irregular path | geometry_recomputed | {
"straightness": 0.46970847600000004,
"max_turn_angle": 33.5774641649,
"closure_ratio": 0.9999653752000001
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "trajectory_shape",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"max_turn_angle",
"closure_ratio"
],
"truth_fields": {
"straightness": 0.46970847600000004,
"max_turn_angle": 33.5774641649,
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_fast.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"straightness": 0.46970847600000004,
"max_turn_angle": 33.5774641649,
"closure_ratio": 0.9999653752000001
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "trajectory_shape",
"reward_kind": "kinematic_composite",
"required_truth_fi... | {"task_metadata":{"straightness":0.4697084760473009,"max_turn_angle":33.57746416488219,"closure_ratio":0.9999653751926174},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"trajectory_shape","reward_kind":"kinematic_composite","required_truth_fields":["straightness","max_turn_angle","closure_r... | <image>
These are frames of a video.
Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. L-shaped path (one major turn)
C. U-shaped / returning path
D. Approximately straight line
Options:
A. Curved / irregular path
B. L-sha... | trajectory_shape | trajectory_shape | true |
B1 | displacement_range | kinematic_perception | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Medium (2 – 5 m)
C. Static (< 0.5 m)
D. Large (> 5 m)
A. Small (0.5 – 2 m)
B. Medium (2 – 5 m)
C. Static (< 0.5 m)
D. Large (> 5 m) | C | {
"A": "Small (0.5 – 2 m)",
"B": "Medium (2 – 5 m)",
"C": "Static (< 0.5 m)",
"D": "Large (> 5 m)"
} | bench_00dd871005_rotation_B1_000 | 0 | null | false | null | true | displacement_range | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Medium (2 – 5 m)
C. Static (< 0.5 m)
D. Large (> 5 m) | Static (< 0.5 m) | geometry_recomputed | {
"displacement_distance": 0
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_distance": 0
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinema... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_distance": 0
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_... | {"task_metadata":{"displacement_distance":0.0},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"displacement_range","reward_kind":"kinematic_scalar","required_truth_fields":["displacement_distance"],"truth_fields":{"displacement_distance":0.0},"apply_answer_reward":true,"apply_format_reward":... | <image>
These are frames of a video.
Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Medium (2 – 5 m)
C. Static (< 0.5 m)
D. Large (> 5 m)
Options:
A. Small (0.5 – 2 m)
B. Medium (2 – 5 m)
C. Static (< 0.5 m)
D. Large ... | displacement_range | displacement_range | true |
B4 | rotation_trap | kinematic_perception | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | The video shows significant visual changes across frames. What is the actual physical displacement of the camera between the first and last frames?
A. Small (0.5 – 2 m)
B. Static (< 0.5 m)
C. Large (> 5 m)
D. Medium (2 – 5 m)
A. Small (0.5 – 2 m)
B. Static (< 0.5 m)
C. Large (> 5 m)
D. Medium (2 – 5 m) | B | {
"A": "Small (0.5 – 2 m)",
"B": "Static (< 0.5 m)",
"C": "Large (> 5 m)",
"D": "Medium (2 – 5 m)"
} | bench_00dd871005_rotation_B4_000 | 0 | null | false | null | true | rotation_trap | The video shows significant visual changes across frames. What is the actual physical displacement of the camera between the first and last frames?
A. Small (0.5 – 2 m)
B. Static (< 0.5 m)
C. Large (> 5 m)
D. Medium (2 – 5 m) | Static (< 0.5 m) | geometry_recomputed | {
"displacement_distance": 0
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_trap",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_distance": 0
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_r... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_distance": 0
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_trap",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_dista... | {"task_metadata":{"displacement_distance":0.0},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"rotation_trap","reward_kind":"kinematic_scalar","required_truth_fields":["displacement_distance"],"truth_fields":{"displacement_distance":0.0},"apply_answer_reward":true,"apply_format_reward":true,... | <image>
These are frames of a video.
The video shows significant visual changes across frames. What is the actual physical displacement of the camera between the first and last frames?
A. Small (0.5 – 2 m)
B. Static (< 0.5 m)
C. Large (> 5 m)
D. Medium (2 – 5 m)
Options:
A. Small (0.5 – 2 m)
B. Static (< 0.5 m)
C. Larg... | rotation_trap | rotation_trap | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | At the beginning of the video, in which direction is the kitchen cabinet relative to the camera? (Directions refer to the horizontal plane)
A. Front-Right
B. Rear-Right
C. Rear-Left
D. Left
A. Front-Right
B. Rear-Right
C. Rear-Left
D. Left | D | {
"A": "Front-Right",
"B": "Rear-Right",
"C": "Rear-Left",
"D": "Left"
} | bench_00dd871005_rotation_B7_000 | "Left" | {
"distance_m": 1.564027448,
"signed_theta_deg": -83.3297766768
} | false | kitchen cabinet | true | first_frame_object_direction | At the beginning of the video, in which direction is the kitchen cabinet relative to the camera? (Directions refer to the horizontal plane)
A. Front-Right
B. Rear-Right
C. Rear-Left
D. Left | Left | geometry_recomputed | {
"object_direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_... | {"task_metadata":{"object_direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Left"},"apply_answer_reward":true,"apply_format_reward... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the kitchen cabinet relative to the camera? (Directions refer to the horizontal plane)
A. Front-Right
B. Rear-Right
C. Rear-Left
D. Left
Options:
A. Front-Right
B. Rear-Right
C. Rear-Left
D. Left
Please answer with the option let... | first_frame_object_direction | first_frame_object_direction | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front
C. Rear
D. Front-Right
A. Left
B. Front
C. Rear
D. Front-Right | B | {
"A": "Left",
"B": "Front",
"C": "Rear",
"D": "Front-Right"
} | bench_00dd871005_rotation_B7_001 | "Front" | {
"distance_m": 2.4489451173,
"signed_theta_deg": 4.955202946
} | false | camping bag | true | first_frame_object_direction | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front
C. Rear
D. Front-Right | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewa... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front
C. Rear
D. Front-Right
Options:
A. Left
B. Front
C. Rear
D. Front-Right
Please answer with the option letter. | first_frame_object_direction | first_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | At the end of the video, in which direction is the kitchen cabinet relative to the camera?
A. Rear-Right
B. Front-Right
C. Left
D. Front-Left
A. Rear-Right
B. Front-Right
C. Left
D. Front-Left | C | {
"A": "Rear-Right",
"B": "Front-Right",
"C": "Left",
"D": "Front-Left"
} | bench_00dd871005_rotation_B8_000 | "Left" | {
"distance_m": 1.5639625105000001,
"signed_theta_deg": -86.3248095553
} | false | kitchen cabinet | true | last_frame_object_direction | At the end of the video, in which direction is the kitchen cabinet relative to the camera?
A. Rear-Right
B. Front-Right
C. Left
D. Front-Left | Left | geometry_recomputed | {
"object_direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_k... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_d... | {"task_metadata":{"object_direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Left"},"apply_answer_reward":true,"apply_format_reward"... | <image>
These are frames of a video.
At the end of the video, in which direction is the kitchen cabinet relative to the camera?
A. Rear-Right
B. Front-Right
C. Left
D. Front-Left
Options:
A. Rear-Right
B. Front-Right
C. Left
D. Front-Left
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | At the end of the video, in which direction is the camping bag relative to the camera?
A. Front
B. Rear-Right
C. Front-Left
D. Right
A. Front
B. Rear-Right
C. Front-Left
D. Right | A | {
"A": "Front",
"B": "Rear-Right",
"C": "Front-Left",
"D": "Right"
} | bench_00dd871005_rotation_B8_003 | "Front" | {
"distance_m": 2.4490740199,
"signed_theta_deg": 1.9225699339000002
} | false | camping bag | true | last_frame_object_direction | At the end of the video, in which direction is the camping bag relative to the camera?
A. Front
B. Rear-Right
C. Front-Left
D. Right | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewar... | <image>
These are frames of a video.
At the end of the video, in which direction is the camping bag relative to the camera?
A. Front
B. Rear-Right
C. Front-Left
D. Right
Options:
A. Front
B. Rear-Right
C. Front-Left
D. Right
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | At the end of the video, in which direction is the toilet paper bag relative to the camera?
A. Front
B. Rear-Right
C. Rear-Left
D. Left
A. Front
B. Rear-Right
C. Rear-Left
D. Left | A | {
"A": "Front",
"B": "Rear-Right",
"C": "Rear-Left",
"D": "Left"
} | bench_00dd871005_rotation_B8_004 | "Front" | {
"distance_m": 2.3799886355,
"signed_theta_deg": 10.5291631397
} | false | toilet paper bag | true | last_frame_object_direction | At the end of the video, in which direction is the toilet paper bag relative to the camera?
A. Front
B. Rear-Right
C. Rear-Left
D. Left | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewar... | <image>
These are frames of a video.
At the end of the video, in which direction is the toilet paper bag relative to the camera?
A. Front
B. Rear-Right
C. Rear-Left
D. Left
Options:
A. Front
B. Rear-Right
C. Rear-Left
D. Left
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | If I am standing by the kitchen cabinet and facing the sofa, is the suitcase to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. back-right
B. front-left
C. back-left
D. front-right | B | {
"A": "back-right",
"B": "front-left",
"C": "back-left",
"D": "front-right"
} | bench_00dd871005_rotation_B11_000 | "front-left" | {
"proj_right": -1.9871164305,
"proj_forward": 2.1320035252,
"signed_theta_deg": -42.9854925157
} | false | kitchen cabinet,sofa,suitcase | true | coordinate_transform_direction | If I am standing by the kitchen cabinet and facing the sofa, is the suitcase to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-left | geometry_recomputed | {
"signed_theta_deg": -42.9854925157,
"relative_direction": "front-left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": -42.9854925157,
"relative_direction": "front-left"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": -42.9854925157,
"relative_direction": "front-left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":-42.98549251573455,"relative_direction":"front-left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sign... | <image>
These are frames of a video.
If I am standing by the kitchen cabinet and facing the sofa, is the suitcase to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. back-right
B. front-... | coordinate_transform_direction | coordinate_transform_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | If I am standing by the kitchen cabinet and facing the sofa, is the camping bag to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. front-right
B. back-left
C. back-right
D. front-left | D | {
"A": "front-right",
"B": "back-left",
"C": "back-right",
"D": "front-left"
} | bench_00dd871005_rotation_B11_001 | "front-left" | {
"proj_right": -2.0757354011,
"proj_forward": 1.5677460236,
"signed_theta_deg": -52.9372494952
} | false | kitchen cabinet,sofa,camping bag | true | coordinate_transform_direction | If I am standing by the kitchen cabinet and facing the sofa, is the camping bag to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-left | geometry_recomputed | {
"signed_theta_deg": -52.9372494952,
"relative_direction": "front-left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": -52.9372494952,
"relative_direction": "front-left"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": -52.9372494952,
"relative_direction": "front-left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":-52.937249495243336,"relative_direction":"front-left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sig... | <image>
These are frames of a video.
If I am standing by the kitchen cabinet and facing the sofa, is the camping bag to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. front-right
B. ba... | coordinate_transform_direction | coordinate_transform_direction | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | Measuring from the closest point of each object, what is the approximate distance between stovetop and sponge?
A. Very close (< 1 m)
B. Close (1 – 3 m)
C. Medium (3 – 6 m)
D. Far (> 6 m) | A | {
"A": "Very close (< 1 m)",
"B": "Close (1 – 3 m)",
"C": "Medium (3 – 6 m)",
"D": "Far (> 6 m)"
} | bench_00dd871005_rotation_B12_000 | 0.8189000000000001 | null | false | stovetop,sponge | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between stovetop and sponge? | Very close (< 1 m) | geometry_recomputed | {
"aabb_surface_distance_m": 0.8189000000000001
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 0.8189000000000001
},
"apply_answer_reward": true,
"apply_format_reward... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 0.8189000000000001
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields"... | {"task_metadata":{"aabb_surface_distance_m":0.8188999999999997},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":0.8188999999999997},"apply_ans... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between stovetop and sponge?
Options:
A. Very close (< 1 m)
B. Close (1 – 3 m)
C. Medium (3 – 6 m)
D. Far (> 6 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | Measuring from the closest point of each object, what is the approximate distance between wardrobe and chair?
A. Medium (3 – 6 m)
B. Far (> 6 m)
C. Close (1 – 3 m)
D. Very close (< 1 m) | A | {
"A": "Medium (3 – 6 m)",
"B": "Far (> 6 m)",
"C": "Close (1 – 3 m)",
"D": "Very close (< 1 m)"
} | bench_00dd871005_rotation_B12_001 | 3.8953 | null | false | wardrobe,chair | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between wardrobe and chair? | Medium (3 – 6 m) | geometry_recomputed | {
"aabb_surface_distance_m": 3.8953
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 3.8953
},
"apply_answer_reward": true,
"apply_format_reward": true,
"... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 3.8953
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"a... | {"task_metadata":{"aabb_surface_distance_m":3.8952999999999998},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":3.8952999999999998},"apply_ans... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between wardrobe and chair?
Options:
A. Medium (3 – 6 m)
B. Far (> 6 m)
C. Close (1 – 3 m)
D. Very close (< 1 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | Measuring from the closest point of each object, what is the approximate distance between paper towel roll and rolling cabinet?
A. Far (> 6 m)
B. Very close (< 1 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m) | D | {
"A": "Far (> 6 m)",
"B": "Very close (< 1 m)",
"C": "Medium (3 – 6 m)",
"D": "Close (1 – 3 m)"
} | bench_00dd871005_rotation_B12_002 | 1.2844902802 | null | false | paper towel roll,rolling cabinet | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between paper towel roll and rolling cabinet? | Close (1 – 3 m) | geometry_recomputed | {
"aabb_surface_distance_m": 1.2844902802
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 1.2844902802
},
"apply_answer_reward": true,
"apply_format_reward": tru... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 1.2844902802
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
... | {"task_metadata":{"aabb_surface_distance_m":1.2844902802279194},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":1.2844902802279194},"apply_ans... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between paper towel roll and rolling cabinet?
Options:
A. Far (> 6 m)
B. Very close (< 1 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B13 | object_count | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | How many window frame objects are visible in this scene?
A. 4
B. 2
C. 3
D. 1
A. 4
B. 2
C. 3
D. 1 | D | {
"A": "4",
"B": "2",
"C": "3",
"D": "1"
} | bench_00dd871005_rotation_B13_000 | 1 | null | false | window frame | true | object_count | How many window frame objects are visible in this scene?
A. 4
B. 2
C. 3
D. 1 | 1 | geometry_recomputed | {
"object_count": 1
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": false,
"prov... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_count": 1
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answe... | {"task_metadata":{"object_count":1},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_count","reward_kind":"spatial_count","required_truth_fields":["object_count"],"truth_fields":{"object_count":1},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic_reward":false,"pro... | <image>
These are frames of a video.
How many window frame objects are visible in this scene?
A. 4
B. 2
C. 3
D. 1
Options:
A. 4
B. 2
C. 3
D. 1
Please answer with the option letter. | object_count | object_count | true |
B14 | object_size | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | What is the longest dimension of the kitchen counter's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Very large (> 2.0 m)
D. Small (< 0.3 m)
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Very large (> 2.0 m)
D. Small (< 0.3 m) | A | {
"A": "Large (1.0 – 2.0 m)",
"B": "Medium (0.3 – 1.0 m)",
"C": "Very large (> 2.0 m)",
"D": "Small (< 0.3 m)"
} | bench_00dd871005_rotation_B14_000 | 1.7429999999999999 | [
1.1455,
1.7429999999999999,
0.11560000000000001
] | false | kitchen counter | true | object_size | What is the longest dimension of the kitchen counter's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Very large (> 2.0 m)
D. Small (< 0.3 m) | Large (1.0 – 2.0 m) | geometry_recomputed | {
"aabb_extents_m": [
1.1455,
1.7429999999999999,
0.11560000000000001
],
"longest_extent_m": 1.7429999999999999
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
1.1455,
1.7429999999999999,
0.11560000000000001
],
"longe... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
1.1455,
1.7429999999999999,
0.11560000000000001
],
"longest_extent_m": 1.7429999999999999
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"require... | {"task_metadata":{"aabb_extents_m":[1.1454999999999997,1.7429999999999999,0.11560000000000004],"longest_extent_m":1.7429999999999999},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth... | <image>
These are frames of a video.
What is the longest dimension of the kitchen counter's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Very large (> 2.0 m)
D. Small (< 0.3 m)
Options:
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Very large (> 2.0 m)
D. Small (< 0.3 m)
Please answer with the o... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | What is the longest dimension of the clothes basket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m) | D | {
"A": "Very large (> 2.0 m)",
"B": "Medium (0.3 – 1.0 m)",
"C": "Large (1.0 – 2.0 m)",
"D": "Small (< 0.3 m)"
} | bench_00dd871005_rotation_B14_001 | 0.2691 | [
0.2169,
0.1913,
0.2691
] | false | clothes basket | true | object_size | What is the longest dimension of the clothes basket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m) | Small (< 0.3 m) | geometry_recomputed | {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
},... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.2169,
0.1913,
0.2691
],
"longest_extent_m": 0.2691
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extent... | {"task_metadata":{"aabb_extents_m":[0.21689999999999987,0.19130000000000003,0.2691],"longest_extent_m":0.2691},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth_fields":{"aabb_extents... | <image>
These are frames of a video.
What is the longest dimension of the clothes basket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
Options:
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
Please answer with the op... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | What is the longest dimension of the blanket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m) | B | {
"A": "Very large (> 2.0 m)",
"B": "Medium (0.3 – 1.0 m)",
"C": "Large (1.0 – 2.0 m)",
"D": "Small (< 0.3 m)"
} | bench_00dd871005_rotation_B14_002 | 0.5632 | [
0.5632,
0.4446,
0.2442
] | false | blanket | true | object_size | What is the longest dimension of the blanket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m) | Medium (0.3 – 1.0 m) | geometry_recomputed | {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
},... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extent... | {"task_metadata":{"aabb_extents_m":[0.5632000000000001,0.4445999999999999,0.24420000000000003],"longest_extent_m":0.5632000000000001},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth... | <image>
These are frames of a video.
What is the longest dimension of the blanket's bounding box?
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
Options:
A. Very large (> 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Large (1.0 – 2.0 m)
D. Small (< 0.3 m)
Please answer with the option le... | object_size | object_size | true |
B17 | motion_type | implicit_trajectory_reasoning | 00dd871005 | rotation | 00dd871005_hq/video_rotation.mp4 | 0.1m | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Stationary rotation (looking around without moving)
B. Static (no significant movement)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction)
A. Stationary rotation (looking aro... | A | {
"A": "Stationary rotation (looking around without moving)",
"B": "Static (no significant movement)",
"C": "Curved path with both translation and rotation",
"D": "Straight-line translation (moving in one direction)"
} | bench_00dd871005_rotation_B17_000 | {
"straightness": 0,
"rotation_deg": 357,
"disp": 0
} | null | false | null | true | motion_type | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Stationary rotation (looking around without moving)
B. Static (no significant movement)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction) | Stationary rotation (looking around without moving) | geometry_recomputed | {
"straightness": 0,
"rotation_deg": 357,
"disp": 0
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"rotation_deg",
"disp"
],
"truth_fields": {
"straightness": 0,
"rotation_deg": 357,
"disp": 0
},
"apply_answer_reward": tr... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_rotation.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"straightness": 0,
"rotation_deg": 357,
"disp": 0
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"rotation_deg",
"d... | {"task_metadata":{"straightness":0.0,"rotation_deg":357.0,"disp":0.0},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"motion_type","reward_kind":"kinematic_composite","required_truth_fields":["straightness","rotation_deg","disp"],"truth_fields":{"straightness":0.0,"rotation_deg":357.0,"disp"... | <image>
These are frames of a video.
Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Stationary rotation (looking around without moving)
B. Static (no significant movement)
C. Curved path with both translation and rotation
D. Straight-line translation (moving in one direction... | motion_type | motion_type | true |
B1 | displacement_range | kinematic_perception | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | C | {
"A": "Small (0.5 – 2 m)",
"B": "Large (> 5 m)",
"C": "Medium (2 – 5 m)",
"D": "Static (< 0.5 m)"
} | bench_00dd871005_slow_B1_000 | 3.0307 | null | false | null | true | displacement_range | Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | Medium (2 – 5 m) | geometry_recomputed | {
"displacement_distance": 3.0307
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displacement_distance": 3.0307
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_k... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_distance": 3.0307
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "displacement_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"displacement_distance"
],
"truth_fields": {
"displace... | {"task_metadata":{"displacement_distance":3.0307},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"displacement_range","reward_kind":"kinematic_scalar","required_truth_fields":["displacement_distance"],"truth_fields":{"displacement_distance":3.0307},"apply_answer_reward":true,"apply_format_re... | <image>
These are frames of a video.
Based on visual cues in the video, what is the straight-line displacement between the camera's starting and ending positions?
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
Options:
A. Small (0.5 – 2 m)
B. Large (> 5 m)
C. Medium (2 – 5 m)
D. Static (<... | displacement_range | displacement_range | true |
B2 | camera_motion_direction | kinematic_perception | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear
B. Front-Right
C. Front
D. Rear-Right
A. Left
B. Right
C. Front-Right
D. Front | A | {
"A": "Left",
"B": "Right",
"C": "Front-Right",
"D": "Front"
} | bench_00dd871005_slow_B2_000 | "Left" | {
"signed_theta_deg": -91.6394929893
} | false | null | true | camera_motion_direction | In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear
B. Front-Right
C. Front
D. Rear-Right | Left | geometry_recomputed | {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "camera_motion_direction",
"reward_kind": "kinematic_direction",
"required_truth_fields": [
"displacement_vector",
"direction"
],
"truth_fields": {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"directi... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"displacement_vector": [
-3.0291,
-0.047,
0.0867
],
"direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "camera_motion_direction",
"reward_kind": "kinematic_direction",
"required_truth_fields": [
... | {"task_metadata":{"displacement_vector":[-3.0291,-0.047,0.0867],"direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"camera_motion_direction","reward_kind":"kinematic_direction","required_truth_fields":["displacement_vector","direction"],"truth_fields":{"displacement_vector":[... | <image>
These are frames of a video.
In which direction did the camera move relative to its starting orientation? (Directions are defined in the horizontal plane)
A. Rear
B. Front-Right
C. Front
D. Rear-Right
Options:
A. Left
B. Right
C. Front-Right
D. Front
Please answer with the option letter. | camera_motion_direction | camera_motion_direction | true |
B3 | rotation_range | kinematic_perception | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Medium (45 – 90°)
B. Large (> 90°)
C. Minimal (< 15°)
D. Small (15 – 45°)
A. Medium (45 – 90°)
B. Large (> 90°)
C. Minimal (< 15°)
D. Small (15 – 45°) | B | {
"A": "Medium (45 – 90°)",
"B": "Large (> 90°)",
"C": "Minimal (< 15°)",
"D": "Small (15 – 45°)"
} | bench_00dd871005_slow_B3_000 | 710.3 | null | false | null | true | rotation_range | What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Medium (45 – 90°)
B. Large (> 90°)
C. Minimal (< 15°)
D. Small (15 – 45°) | Large (> 90°) | geometry_recomputed | {
"rotation_angle": 710.3
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"rotation_angle"
],
"truth_fields": {
"rotation_angle": 710.3
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": t... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"rotation_angle": 710.3
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "rotation_range",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"rotation_angle"
],
"truth_fields": {
"rotation_angle": 710.3
... | {"task_metadata":{"rotation_angle":710.3},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"rotation_range","reward_kind":"kinematic_scalar","required_truth_fields":["rotation_angle"],"truth_fields":{"rotation_angle":710.3},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic... | <image>
These are frames of a video.
What is the approximate rotation angle between the camera's orientation in the first frame and the last frame?
A. Medium (45 – 90°)
B. Large (> 90°)
C. Minimal (< 15°)
D. Small (15 – 45°)
Options:
A. Medium (45 – 90°)
B. Large (> 90°)
C. Minimal (< 15°)
D. Small (15 – 45°)
Please a... | rotation_range | rotation_range | true |
B6 | average_speed | kinematic_perception | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | What is the approximate average movement speed of the camera throughout this video?
A. Fast (> 3.0 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Slow (0.5 – 1.5 m/s)
D. Moderate (1.5 – 3.0 m/s)
A. Fast (> 3.0 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Slow (0.5 – 1.5 m/s)
D. Moderate (1.5 – 3.0 m/s) | B | {
"A": "Fast (> 3.0 m/s)",
"B": "Nearly stationary (< 0.5 m/s)",
"C": "Slow (0.5 – 1.5 m/s)",
"D": "Moderate (1.5 – 3.0 m/s)"
} | bench_00dd871005_slow_B6_000 | 0.21517000000000003 | null | false | null | true | average_speed | What is the approximate average movement speed of the camera throughout this video?
A. Fast (> 3.0 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Slow (0.5 – 1.5 m/s)
D. Moderate (1.5 – 3.0 m/s) | Nearly stationary (< 0.5 m/s) | geometry_recomputed | {
"average_speed": 0.21517000000000003
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "average_speed",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"average_speed"
],
"truth_fields": {
"average_speed": 0.21517000000000003
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"average_speed": 0.21517000000000003
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "average_speed",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"average_speed"
],
"truth_fields": {
"average_speed": ... | {"task_metadata":{"average_speed":0.21517},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"average_speed","reward_kind":"kinematic_scalar","required_truth_fields":["average_speed"],"truth_fields":{"average_speed":0.21517},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic... | <image>
These are frames of a video.
What is the approximate average movement speed of the camera throughout this video?
A. Fast (> 3.0 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Slow (0.5 – 1.5 m/s)
D. Moderate (1.5 – 3.0 m/s)
Options:
A. Fast (> 3.0 m/s)
B. Nearly stationary (< 0.5 m/s)
C. Slow (0.5 – 1.5 m/s)
D. Moder... | average_speed | average_speed | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front-Left
C. Rear
D. Rear-Right
A. Rear-Left
B. Front-Left
C. Rear
D. Rear-Right | B | {
"A": "Rear-Left",
"B": "Front-Left",
"C": "Rear",
"D": "Rear-Right"
} | bench_00dd871005_slow_B7_001 | "Front-Left" | {
"distance_m": 1.0950316998,
"signed_theta_deg": -47.5338735644
} | false | camping bag | true | first_frame_object_direction | At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front-Left
C. Rear
D. Rear-Right | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"o... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_f... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the camping bag relative to the camera? (Directions refer to the horizontal plane)
A. Rear-Left
B. Front-Left
C. Rear
D. Rear-Right
Options:
A. Rear-Left
B. Front-Left
C. Rear
D. Rear-Right
Please answer with the option letter. | first_frame_object_direction | first_frame_object_direction | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the beginning of the video, in which direction is the reusable shopping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front-Left
C. Rear
D. Front
A. Left
B. Front-Left
C. Rear
D. Front | D | {
"A": "Left",
"B": "Front-Left",
"C": "Rear",
"D": "Front"
} | bench_00dd871005_slow_B7_002 | "Front" | {
"distance_m": 1.6139739337,
"signed_theta_deg": -9.4255677329
} | false | reusable shopping bag | true | first_frame_object_direction | At the beginning of the video, in which direction is the reusable shopping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front-Left
C. Rear
D. Front | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewa... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the reusable shopping bag relative to the camera? (Directions refer to the horizontal plane)
A. Left
B. Front-Left
C. Rear
D. Front
Options:
A. Left
B. Front-Left
C. Rear
D. Front
Please answer with the option letter. | first_frame_object_direction | first_frame_object_direction | true |
B7 | first_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Right
B. Rear-Right
C. Front
D. Front-Right
A. Right
B. Rear-Right
C. Front
D. Front-Right | C | {
"A": "Right",
"B": "Rear-Right",
"C": "Front",
"D": "Front-Right"
} | bench_00dd871005_slow_B7_003 | "Front" | {
"distance_m": 1.7478991375000001,
"signed_theta_deg": 8.7110567842
} | false | clothes | true | first_frame_object_direction | At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Right
B. Rear-Right
C. Front
D. Front-Right | Front | geometry_recomputed | {
"object_direction": "Front"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "first_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object... | {"task_metadata":{"object_direction":"Front"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"first_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front"},"apply_answer_reward":true,"apply_format_rewa... | <image>
These are frames of a video.
At the beginning of the video, in which direction is the clothes relative to the camera? (Directions refer to the horizontal plane)
A. Right
B. Rear-Right
C. Front
D. Front-Right
Options:
A. Right
B. Rear-Right
C. Front
D. Front-Right
Please answer with the option letter. | first_frame_object_direction | first_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the end of the video, in which direction is the sofa relative to the camera?
A. Left
B. Front
C. Rear-Right
D. Rear
A. Left
B. Front
C. Rear-Right
D. Rear | A | {
"A": "Left",
"B": "Front",
"C": "Rear-Right",
"D": "Rear"
} | bench_00dd871005_slow_B8_001 | "Left" | {
"distance_m": 2.623024984,
"signed_theta_deg": -79.7868636093
} | false | sofa | true | last_frame_object_direction | At the end of the video, in which direction is the sofa relative to the camera?
A. Left
B. Front
C. Rear-Right
D. Rear | Left | geometry_recomputed | {
"object_direction": "Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_k... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_d... | {"task_metadata":{"object_direction":"Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Left"},"apply_answer_reward":true,"apply_format_reward"... | <image>
These are frames of a video.
At the end of the video, in which direction is the sofa relative to the camera?
A. Left
B. Front
C. Rear-Right
D. Rear
Options:
A. Left
B. Front
C. Rear-Right
D. Rear
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Rear-Left
C. Front
D. Rear-Right
A. Front-Left
B. Rear-Left
C. Front
D. Rear-Right | A | {
"A": "Front-Left",
"B": "Rear-Left",
"C": "Front",
"D": "Rear-Right"
} | bench_00dd871005_slow_B8_003 | "Front-Left" | {
"distance_m": 2.5160079631,
"signed_theta_deg": -40.2457799648
} | false | power sockets | true | last_frame_object_direction | At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Rear-Left
C. Front
D. Rear-Right | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"a... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"ob... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_fo... | <image>
These are frames of a video.
At the end of the video, in which direction is the power sockets relative to the camera?
A. Front-Left
B. Rear-Left
C. Front
D. Rear-Right
Options:
A. Front-Left
B. Rear-Left
C. Front
D. Rear-Right
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B8 | last_frame_object_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | At the end of the video, in which direction is the phone charger relative to the camera?
A. Front
B. Rear
C. Front-Left
D. Left
A. Front
B. Rear
C. Front-Left
D. Left | C | {
"A": "Front",
"B": "Rear",
"C": "Front-Left",
"D": "Left"
} | bench_00dd871005_slow_B8_004 | "Front-Left" | {
"distance_m": 2.4899383095,
"signed_theta_deg": -41.0543956034
} | false | phone charger | true | last_frame_object_direction | At the end of the video, in which direction is the phone charger relative to the camera?
A. Front
B. Rear
C. Front-Left
D. Left | Front-Left | geometry_recomputed | {
"object_direction": "Front-Left"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"object_direction": "Front-Left"
},
"apply_answer_reward": true,
"apply_format_reward": true,
"a... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_direction": "Front-Left"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "last_frame_object_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"object_direction"
],
"truth_fields": {
"ob... | {"task_metadata":{"object_direction":"Front-Left"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"last_frame_object_direction","reward_kind":"spatial_direction","required_truth_fields":["object_direction"],"truth_fields":{"object_direction":"Front-Left"},"apply_answer_reward":true,"apply_fo... | <image>
These are frames of a video.
At the end of the video, in which direction is the phone charger relative to the camera?
A. Front
B. Rear
C. Front-Left
D. Left
Options:
A. Front
B. Rear
C. Front-Left
D. Left
Please answer with the option letter. | last_frame_object_direction | last_frame_object_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. front-right
B. back-left
C. front-left
D. back-right | A | {
"A": "front-right",
"B": "back-left",
"C": "front-left",
"D": "back-right"
} | bench_00dd871005_slow_B11_000 | "front-right" | {
"proj_right": 1.1205207594,
"proj_forward": 0.6019209897000001,
"signed_theta_deg": 61.7562112133
} | false | bed,sofa,phone charger | true | coordinate_transform_direction | If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-right | geometry_recomputed | {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": 61.7562112133,
"relative_direction": "front-right"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":61.75621121325139,"relative_direction":"front-right"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sign... | <image>
These are frames of a video.
If I am standing by the bed and facing the sofa, is the phone charger to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. front-right
B. back-left
C.... | coordinate_transform_direction | coordinate_transform_direction | true |
B11 | coordinate_transform_direction | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
A. front-left
B. front-right
C. back-left
D. back-right | B | {
"A": "front-left",
"B": "front-right",
"C": "back-left",
"D": "back-right"
} | bench_00dd871005_slow_B11_001 | "front-right" | {
"proj_right": 0.9021692509,
"proj_forward": 0.9590482901,
"signed_theta_deg": 43.2495786434
} | false | bed,sofa,projector | true | coordinate_transform_direction | If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis). | front-right | geometry_recomputed | {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_theta_deg",
"relative_direction"
],
"truth_fields": {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"signed_theta_deg": 43.2495786434,
"relative_direction": "front-right"
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "coordinate_transform_direction",
"reward_kind": "spatial_direction",
"required_truth_fields": [
"signed_thet... | {"task_metadata":{"signed_theta_deg":43.249578643434816,"relative_direction":"front-right"},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"coordinate_transform_direction","reward_kind":"spatial_direction","required_truth_fields":["signed_theta_deg","relative_direction"],"truth_fields":{"sig... | <image>
These are frames of a video.
If I am standing by the bed and facing the sofa, is the projector to my front-left, front-right, back-left, or back-right? Directions refer to the quadrants of a Cartesian plane (assuming I am at the origin and facing the positive y-axis).
Options:
A. front-left
B. front-right
C. ba... | coordinate_transform_direction | coordinate_transform_direction | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Measuring from the closest point of each object, what is the approximate distance between boxes and jar?
A. Close (1 – 3 m)
B. Far (> 6 m)
C. Medium (3 – 6 m)
D. Very close (< 1 m) | D | {
"A": "Close (1 – 3 m)",
"B": "Far (> 6 m)",
"C": "Medium (3 – 6 m)",
"D": "Very close (< 1 m)"
} | bench_00dd871005_slow_B12_000 | 0.0176 | null | false | boxes,jar | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between boxes and jar? | Very close (< 1 m) | geometry_recomputed | {
"aabb_surface_distance_m": 0.0176
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 0.0176
},
"apply_answer_reward": true,
"apply_format_reward": true,
"... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 0.0176
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"a... | {"task_metadata":{"aabb_surface_distance_m":0.017600000000000282},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":0.017600000000000282},"apply... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between boxes and jar?
Options:
A. Close (1 – 3 m)
B. Far (> 6 m)
C. Medium (3 – 6 m)
D. Very close (< 1 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Measuring from the closest point of each object, what is the approximate distance between oranges and sponge?
A. Far (> 6 m)
B. Very close (< 1 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m) | D | {
"A": "Far (> 6 m)",
"B": "Very close (< 1 m)",
"C": "Medium (3 – 6 m)",
"D": "Close (1 – 3 m)"
} | bench_00dd871005_slow_B12_001 | 1.2846210725 | null | false | oranges,sponge | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between oranges and sponge? | Close (1 – 3 m) | geometry_recomputed | {
"aabb_surface_distance_m": 1.2846210725
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 1.2846210725
},
"apply_answer_reward": true,
"apply_format_reward": tru... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 1.2846210725
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
... | {"task_metadata":{"aabb_surface_distance_m":1.2846210725346212},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":1.2846210725346212},"apply_ans... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between oranges and sponge?
Options:
A. Far (> 6 m)
B. Very close (< 1 m)
C. Medium (3 – 6 m)
D. Close (1 – 3 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B12 | inter_object_distance | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Measuring from the closest point of each object, what is the approximate distance between microwave and jar?
A. Very close (< 1 m)
B. Medium (3 – 6 m)
C. Close (1 – 3 m)
D. Far (> 6 m) | A | {
"A": "Very close (< 1 m)",
"B": "Medium (3 – 6 m)",
"C": "Close (1 – 3 m)",
"D": "Far (> 6 m)"
} | bench_00dd871005_slow_B12_002 | 0.4792065212 | null | false | microwave,jar | true | inter_object_distance | Measuring from the closest point of each object, what is the approximate distance between microwave and jar? | Very close (< 1 m) | geometry_recomputed | {
"aabb_surface_distance_m": 0.4792065212
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
"aabb_surface_distance_m": 0.4792065212
},
"apply_answer_reward": true,
"apply_format_reward": tru... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_surface_distance_m": 0.4792065212
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "inter_object_distance",
"reward_kind": "spatial_distance",
"required_truth_fields": [
"aabb_surface_distance_m"
],
"truth_fields": {
... | {"task_metadata":{"aabb_surface_distance_m":0.47920652124110336},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"inter_object_distance","reward_kind":"spatial_distance","required_truth_fields":["aabb_surface_distance_m"],"truth_fields":{"aabb_surface_distance_m":0.47920652124110336},"apply_a... | <image>
These are frames of a video.
Measuring from the closest point of each object, what is the approximate distance between microwave and jar?
Options:
A. Very close (< 1 m)
B. Medium (3 – 6 m)
C. Close (1 – 3 m)
D. Far (> 6 m)
Please answer with the option letter. | inter_object_distance | inter_object_distance | true |
B13 | object_count | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | How many sofa objects are visible in this scene?
A. 2
B. 3
C. 4
D. 1
A. 2
B. 3
C. 4
D. 1 | D | {
"A": "2",
"B": "3",
"C": "4",
"D": "1"
} | bench_00dd871005_slow_B13_000 | 1 | null | false | sofa | true | object_count | How many sofa objects are visible in this scene?
A. 2
B. 3
C. 4
D. 1 | 1 | geometry_recomputed | {
"object_count": 1
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": false,
"prov... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_count": 1
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answe... | {"task_metadata":{"object_count":1},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_count","reward_kind":"spatial_count","required_truth_fields":["object_count"],"truth_fields":{"object_count":1},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic_reward":false,"pro... | <image>
These are frames of a video.
How many sofa objects are visible in this scene?
A. 2
B. 3
C. 4
D. 1
Options:
A. 2
B. 3
C. 4
D. 1
Please answer with the option letter. | object_count | object_count | true |
B13 | object_count | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | How many reusable shopping bag objects are visible in this scene?
A. 4
B. 3
C. 2
D. 1
A. 4
B. 3
C. 2
D. 1 | D | {
"A": "4",
"B": "3",
"C": "2",
"D": "1"
} | bench_00dd871005_slow_B13_002 | 1 | null | false | reusable shopping bag | true | object_count | How many reusable shopping bag objects are visible in this scene?
A. 4
B. 3
C. 2
D. 1 | 1 | geometry_recomputed | {
"object_count": 1
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": false,
"prov... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"object_count": 1
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_count",
"reward_kind": "spatial_count",
"required_truth_fields": [
"object_count"
],
"truth_fields": {
"object_count": 1
},
"apply_answe... | {"task_metadata":{"object_count":1},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_count","reward_kind":"spatial_count","required_truth_fields":["object_count"],"truth_fields":{"object_count":1},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinematic_reward":false,"pro... | <image>
These are frames of a video.
How many reusable shopping bag objects are visible in this scene?
A. 4
B. 3
C. 2
D. 1
Options:
A. 4
B. 3
C. 2
D. 1
Please answer with the option letter. | object_count | object_count | true |
B14 | object_size | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | What is the longest dimension of the blanket's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | B | {
"A": "Large (1.0 – 2.0 m)",
"B": "Medium (0.3 – 1.0 m)",
"C": "Small (< 0.3 m)",
"D": "Very large (> 2.0 m)"
} | bench_00dd871005_slow_B14_000 | 0.5632 | [
0.5632,
0.4446,
0.2442
] | false | blanket | true | object_size | What is the longest dimension of the blanket's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | Medium (0.3 – 1.0 m) | geometry_recomputed | {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
},... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.5632,
0.4446,
0.2442
],
"longest_extent_m": 0.5632
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extent... | {"task_metadata":{"aabb_extents_m":[0.5632000000000001,0.4445999999999999,0.24420000000000003],"longest_extent_m":0.5632000000000001},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"truth... | <image>
These are frames of a video.
What is the longest dimension of the blanket's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Options:
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Please answer with the option le... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | What is the longest dimension of the oranges's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | C | {
"A": "Large (1.0 – 2.0 m)",
"B": "Medium (0.3 – 1.0 m)",
"C": "Small (< 0.3 m)",
"D": "Very large (> 2.0 m)"
} | bench_00dd871005_slow_B14_001 | 0.1358 | [
0.1358,
0.125,
0.1136
] | false | oranges | true | object_size | What is the longest dimension of the oranges's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m) | Small (< 0.3 m) | geometry_recomputed | {
"aabb_extents_m": [
0.1358,
0.125,
0.1136
],
"longest_extent_m": 0.1358
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.1358,
0.125,
0.1136
],
"longest_extent_m": 0.1358
},
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.1358,
0.125,
0.1136
],
"longest_extent_m": 0.1358
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents... | {"task_metadata":{"aabb_extents_m":[0.13579999999999992,0.12499999999999956,0.11359999999999992],"longest_extent_m":0.13579999999999992},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"tr... | <image>
These are frames of a video.
What is the longest dimension of the oranges's bounding box?
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Options:
A. Large (1.0 – 2.0 m)
B. Medium (0.3 – 1.0 m)
C. Small (< 0.3 m)
D. Very large (> 2.0 m)
Please answer with the option le... | object_size | object_size | true |
B14 | object_size | spatial_understanding | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | What is the longest dimension of the toilet paper bag's bounding box?
A. Large (1.0 – 2.0 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Small (< 0.3 m)
A. Large (1.0 – 2.0 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Small (< 0.3 m) | D | {
"A": "Large (1.0 – 2.0 m)",
"B": "Very large (> 2.0 m)",
"C": "Medium (0.3 – 1.0 m)",
"D": "Small (< 0.3 m)"
} | bench_00dd871005_slow_B14_002 | 0.265 | [
0.16340000000000002,
0.1738,
0.265
] | false | toilet paper bag | true | object_size | What is the longest dimension of the toilet paper bag's bounding box?
A. Large (1.0 – 2.0 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Small (< 0.3 m) | Small (< 0.3 m) | geometry_recomputed | {
"aabb_extents_m": [
0.16340000000000002,
0.1738,
0.265
],
"longest_extent_m": 0.265
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"aabb_extents_m",
"longest_extent_m"
],
"truth_fields": {
"aabb_extents_m": [
0.16340000000000002,
0.1738,
0.265
],
"longest_extent_m":... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"aabb_extents_m": [
0.16340000000000002,
0.1738,
0.265
],
"longest_extent_m": 0.265
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "object_size",
"reward_kind": "spatial_size",
"required_truth_fields": [
"... | {"task_metadata":{"aabb_extents_m":[0.16339999999999932,0.17379999999999995,0.2650000000000001],"longest_extent_m":0.2650000000000001},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"object_size","reward_kind":"spatial_size","required_truth_fields":["aabb_extents_m","longest_extent_m"],"trut... | <image>
These are frames of a video.
What is the longest dimension of the toilet paper bag's bounding box?
A. Large (1.0 – 2.0 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Small (< 0.3 m)
Options:
A. Large (1.0 – 2.0 m)
B. Very large (> 2.0 m)
C. Medium (0.3 – 1.0 m)
D. Small (< 0.3 m)
Please answer with the ... | object_size | object_size | true |
B17 | motion_type | implicit_trajectory_reasoning | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Straight-line translation (moving in one direction)
C. Curved path with both translation and rotation
D. Stationary rotation (looking around without moving)
A. Static (no significant movement)... | C | {
"A": "Static (no significant movement)",
"B": "Straight-line translation (moving in one direction)",
"C": "Curved path with both translation and rotation",
"D": "Stationary rotation (looking around without moving)"
} | bench_00dd871005_slow_B17_000 | {
"straightness": 0.46950473270000004,
"rotation_deg": 710.3,
"disp": 3.0307
} | null | false | null | true | motion_type | Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Straight-line translation (moving in one direction)
C. Curved path with both translation and rotation
D. Stationary rotation (looking around without moving) | Curved path with both translation and rotation | geometry_recomputed | {
"straightness": 0.46950473270000004,
"rotation_deg": 710.3,
"disp": 3.0307
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"rotation_deg",
"disp"
],
"truth_fields": {
"straightness": 0.46950473270000004,
"rotation_deg": 710.3,
"disp": 3.0307
},
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"straightness": 0.46950473270000004,
"rotation_deg": 710.3,
"disp": 3.0307
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "motion_type",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
... | {"task_metadata":{"straightness":0.46950473269198,"rotation_deg":710.3,"disp":3.0307},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"motion_type","reward_kind":"kinematic_composite","required_truth_fields":["straightness","rotation_deg","disp"],"truth_fields":{"straightness":0.4695047326919... | <image>
These are frames of a video.
Based on the visual cues in the video, which best describes the camera's motion pattern?
A. Static (no significant movement)
B. Straight-line translation (moving in one direction)
C. Curved path with both translation and rotation
D. Stationary rotation (looking around without moving... | motion_type | motion_type | true |
B18 | trajectory_shape | implicit_trajectory_reasoning | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. Approximately straight line
C. L-shaped path (one major turn)
D. U-shaped / returning path
A. Curved / irregular path
B. Approximately straight line
C. L-shaped path (one m... | A | {
"A": "Curved / irregular path",
"B": "Approximately straight line",
"C": "L-shaped path (one major turn)",
"D": "U-shaped / returning path"
} | bench_00dd871005_slow_B18_000 | {
"straightness": 0.46950473270000004,
"max_turn_angle": 0.9452758898,
"closure_ratio": 0.9999653752000001
} | null | false | null | true | trajectory_shape | Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. Approximately straight line
C. L-shaped path (one major turn)
D. U-shaped / returning path | Curved / irregular path | geometry_recomputed | {
"straightness": 0.46950473270000004,
"max_turn_angle": 0.9452758898,
"closure_ratio": 0.9999653752000001
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "trajectory_shape",
"reward_kind": "kinematic_composite",
"required_truth_fields": [
"straightness",
"max_turn_angle",
"closure_ratio"
],
"truth_fields": {
"straightness": 0.46950473270000004,
"max_turn_angle": 0.9452758898,
... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"straightness": 0.46950473270000004,
"max_turn_angle": 0.9452758898,
"closure_ratio": 0.9999653752000001
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "trajectory_shape",
"reward_kind": "kinematic_composite",
"required_truth_fie... | {"task_metadata":{"straightness":0.46950473269198,"max_turn_angle":0.9452758897942858,"closure_ratio":0.9999653751926174},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"trajectory_shape","reward_kind":"kinematic_composite","required_truth_fields":["straightness","max_turn_angle","closure_ra... | <image>
These are frames of a video.
Based on the visual changes throughout the video, which best describes the shape of the camera's movement trajectory?
A. Curved / irregular path
B. Approximately straight line
C. L-shaped path (one major turn)
D. U-shaped / returning path
Options:
A. Curved / irregular path
B. Appro... | trajectory_shape | trajectory_shape | true |
B20 | total_distance_traveled | kinematic_perception | 00dd871005 | slow | 00dd871005_hq/video_slow.mp4 | 2.0m | Based on visual cues in the video, how much total distance did the camera move?
A. Large (> 5 m)
B. Small (0.5 – 2 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
A. Large (> 5 m)
B. Small (0.5 – 2 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | A | {
"A": "Large (> 5 m)",
"B": "Small (0.5 – 2 m)",
"C": "Medium (2 – 5 m)",
"D": "Static (< 0.5 m)"
} | bench_00dd871005_slow_B20_000 | 6.4551 | null | false | null | true | total_distance | Based on visual cues in the video, how much total distance did the camera move?
A. Large (> 5 m)
B. Small (0.5 – 2 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m) | Large (> 5 m) | geometry_recomputed | {
"total_distance": 6.4551
} | {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "total_distance",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"total_distance"
],
"truth_fields": {
"total_distance": 6.4551
},
"apply_answer_reward": true,
"apply_format_reward": true,
"apply_kinematic_reward": ... | {
"schema_version": "legacy_vsi_aligned_v1",
"canonical_video_json": "/root/scannetpp_work/nerfstudio_outputs/00dd871005_hq/video_slow.json",
"reference_parameters": "cold_pipeline628",
"recomputed": true
} | {
"task_metadata": {
"total_distance": 6.4551
},
"reward_targets": {
"schema_version": "legacy_vsi_aligned_v1",
"question_type": "total_distance",
"reward_kind": "kinematic_scalar",
"required_truth_fields": [
"total_distance"
],
"truth_fields": {
"total_distance": 6.4551
... | {"task_metadata":{"total_distance":6.4551},"reward_targets":{"schema_version":"legacy_vsi_aligned_v1","question_type":"total_distance","reward_kind":"kinematic_scalar","required_truth_fields":["total_distance"],"truth_fields":{"total_distance":6.4551},"apply_answer_reward":true,"apply_format_reward":true,"apply_kinemat... | <image>
These are frames of a video.
Based on visual cues in the video, how much total distance did the camera move?
A. Large (> 5 m)
B. Small (0.5 – 2 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
Options:
A. Large (> 5 m)
B. Small (0.5 – 2 m)
C. Medium (2 – 5 m)
D. Static (< 0.5 m)
Please answer with the option letter. | total_distance | total_distance | true |
End of preview. Expand in Data Studio
Dyn3D Benchmark
This release contains the existing final Dyn3D evaluation annotations and their corresponding rendered videos: 16,063 questions, 167 scenes, 835 videos, and 19 question types.
Each scene has five camera trajectories: fast, slow, smooth, rotation, and translation. This release contains the benchmark evaluation split. SFT and RL training data and WorldCam-Bench are separate datasets and are not included.
Files
benchmark.jsonl: original final annotations, preserved byte for byte.videos/{scene_id}_hq/video_{trajectory}.mp4: the 835 referenced videos.release-manifest.json: release counts, file sizes, and SHA-256 hashes.
The annotation video field is relative to the videos/ directory. For example,
00dd871005_hq/video_fast.mp4 resolves to
videos/00dd871005_hq/video_fast.mp4 in this repository.
Reading the original annotations
import json
from pathlib import Path
dataset_root = Path("/path/to/downloaded/repository")
with (dataset_root / "benchmark.jsonl").open(encoding="utf-8") as handle:
for line in handle:
row = json.loads(line)
video_path = dataset_root / "videos" / row["video"]
question = row["question"]
options = row["options"]
answer = row["correct_answer"]
The original annotations also retain geometric ground truth, alignment
provenance, and evaluation metadata. This upload does not rewrite questions,
options, or answers and does not constitute a new correctness audit.
All 16,063 records carry train_use_forbidden: true.
- Downloads last month
- 136