diff --git "a/data/vsibench_test/qa_mobj_3dBbox_json_conversation.json" "b/data/vsibench_test/qa_mobj_3dBbox_json_conversation.json" deleted file mode 100644--- "a/data/vsibench_test/qa_mobj_3dBbox_json_conversation.json" +++ /dev/null @@ -1,10498 +0,0 @@ -[ - { - "id": 0, - "data_source": "Mono3DRefer", - "scene_name": "000002", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[657.52, 217.65], [688.67, 217.63], [700.28, 223.7], [664.91, 223.72], [657.52, 189.82], [688.67, 189.82], [700.28, 192.11], [664.91, 192.12]]}]\n```" - } - ] - }, - { - "id": 1, - "data_source": "Mono3DRefer", - "scene_name": "000004", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[315.67, 215.83], [280.63, 215.83], [313.33, 211.55], [344.88, 211.55], [315.67, 186.18], [280.63, 186.18], [313.33, 184.85], [344.88, 184.85]]}, {\"category\": \"car\", \"corners_3d\": [[391.52, 205.38], [365.35, 205.37], [381.52, 203.27], [406.01, 203.28], [391.52, 185.25], [365.35, 185.24], [381.52, 184.45], [406.01, 184.45]]}]\n```" - } - ] - }, - { - "id": 2, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[602.7, 187.07], [627.8, 187.07], [629.84, 189.85], [599.85, 189.84], [602.7, 159.88], [627.8, 159.87], [629.84, 157.34], [599.85, 157.34]]}]\n```" - } - ] - }, - { - "id": 3, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[676.86, 193.17], [686.12, 193.18], [688.89, 194.1], [679.22, 194.09], [676.86, 164.53], [686.12, 164.53], [688.89, 164.16], [679.22, 164.16]]}]\n```" - } - ] - }, - { - "id": 4, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[411.71, 203.29], [387.88, 203.29], [401.4, 201.43], [423.77, 201.43], [411.71, 182.02], [387.88, 182.02], [401.4, 181.46], [423.77, 181.46]]}]\n```" - } - ] - }, - { - "id": 5, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[602.7, 187.07], [627.8, 187.07], [629.84, 189.85], [599.85, 189.84], [602.7, 159.88], [627.8, 159.87], [629.84, 157.34], [599.85, 157.34]]}, {\"category\": \"cyclist\", \"corners_3d\": [[676.86, 193.17], [686.12, 193.18], [688.89, 194.1], [679.22, 194.09], [676.86, 164.53], [686.12, 164.53], [688.89, 164.16], [679.22, 164.16]]}]\n```" - } - ] - }, - { - "id": 6, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[602.7, 187.07], [627.8, 187.07], [629.84, 189.85], [599.85, 189.84], [602.7, 159.88], [627.8, 159.87], [629.84, 157.34], [599.85, 157.34]]}, {\"category\": \"car\", \"corners_3d\": [[411.71, 203.29], [387.88, 203.29], [401.4, 201.43], [423.77, 201.43], [411.71, 182.02], [387.88, 182.02], [401.4, 181.46], [423.77, 181.46]]}]\n```" - } - ] - }, - { - "id": 7, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[411.71, 203.29], [387.88, 203.29], [401.4, 201.43], [423.77, 201.43], [411.71, 182.02], [387.88, 182.02], [401.4, 181.46], [423.77, 181.46]]}, {\"category\": \"cyclist\", \"corners_3d\": [[676.86, 193.17], [686.12, 193.18], [688.89, 194.1], [679.22, 194.09], [676.86, 164.53], [686.12, 164.53], [688.89, 164.16], [679.22, 164.16]]}]\n```" - } - ] - }, - { - "id": 8, - "data_source": "Mono3DRefer", - "scene_name": "000001", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[602.7, 187.07], [627.8, 187.07], [629.84, 189.85], [599.85, 189.84], [602.7, 159.88], [627.8, 159.87], [629.84, 157.34], [599.85, 157.34]]}, {\"category\": \"car\", \"corners_3d\": [[411.71, 203.29], [387.88, 203.29], [401.4, 201.43], [423.77, 201.43], [411.71, 182.02], [387.88, 182.02], [401.4, 181.46], [423.77, 181.46]]}, {\"category\": \"cyclist\", \"corners_3d\": [[676.86, 193.17], [686.12, 193.18], [688.89, 194.1], [679.22, 194.09], [676.86, 164.53], [686.12, 164.53], [688.89, 164.16], [679.22, 164.16]]}]\n```" - } - ] - }, - { - "id": 9, - "data_source": "Mono3DRefer", - "scene_name": "000005", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[356.21, 238.67], [325.92, 238.62], [334.19, 236.79], [363.65, 236.84], [356.21, 179.2], [325.92, 179.2], [334.19, 179.02], [363.65, 179.03]]}]\n```" - } - ] - }, - { - "id": 10, - "data_source": "Mono3DRefer", - "scene_name": "000037", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[413.23, 244.01], [510.88, 243.86], [-147.67, 711.93], [-1059.05, 724.5], [413.23, 86.2], [510.88, 86.45], [-147.67, -701.84], [-1059.05, -723.01]]}]\n```" - } - ] - }, - { - "id": 11, - "data_source": "Mono3DRefer", - "scene_name": "000037", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[563.17, 229.65], [622.43, 229.58], [628.79, 239.3], [556.53, 239.4], [563.17, 175.74], [622.43, 175.76], [628.79, 173.69], [556.53, 173.67]]}, {\"category\": \"car\", \"corners_3d\": [[492.11, 218.61], [535.62, 218.56], [525.68, 225.11], [473.57, 225.19], [492.11, 183.57], [535.62, 183.57], [525.68, 183.25], [473.57, 183.24]]}]\n```" - } - ] - }, - { - "id": 12, - "data_source": "Mono3DRefer", - "scene_name": "000037", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[563.17, 229.65], [622.43, 229.58], [628.79, 239.3], [556.53, 239.4], [563.17, 175.74], [622.43, 175.76], [628.79, 173.69], [556.53, 173.67]]}, {\"category\": \"truck\", \"corners_3d\": [[413.23, 244.01], [510.88, 243.86], [-147.67, 711.93], [-1059.05, 724.5], [413.23, 86.2], [510.88, 86.45], [-147.67, -701.84], [-1059.05, -723.01]]}, {\"category\": \"car\", \"corners_3d\": [[492.11, 218.61], [535.62, 218.56], [525.68, 225.11], [473.57, 225.19], [492.11, 183.57], [535.62, 183.57], [525.68, 183.25], [473.57, 183.24]]}]\n```" - } - ] - }, - { - "id": 13, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[404.08, 264.55], [490.02, 264.23], [453.7, 297.94], [335.73, 298.54], [404.08, 180.01], [490.02, 179.98], [453.7, 182.61], [335.73, 182.66]]}, {\"category\": \"car\", \"corners_3d\": [[786.75, 289.02], [894.88, 288.68], [1028.73, 340.51], [872.58, 341.21], [786.75, 179.32], [894.88, 179.3], [1028.73, 182.19], [872.58, 182.23]]}, {\"category\": \"car\", \"corners_3d\": [[713.05, 253.26], [789.23, 253.1], [849.18, 277.96], [749.5, 278.24], [713.05, 179.14], [789.23, 179.12], [849.18, 181.06], [749.5, 181.09]]}, {\"category\": \"car\", \"corners_3d\": [[661.26, 215.95], [702.21, 216.02], [712.86, 222.55], [665.71, 222.45], [661.26, 178.71], [702.21, 178.72], [712.86, 179.61], [665.71, 179.59]]}]\n```" - } - ] - }, - { - "id": 14, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[522.53, 228.87], [446.41, 228.93], [480.13, 216.76], [539.71, 216.72], [522.53, 133.03], [446.41, 132.98], [480.13, 141.64], [539.71, 141.67]]}]\n```" - } - ] - }, - { - "id": 15, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[621.29, 198.32], [645.55, 198.32], [649.33, 200.57], [622.92, 200.58], [621.29, 172.05], [645.55, 172.05], [649.33, 171.98], [622.92, 171.98]]}]\n```" - } - ] - }, - { - "id": 16, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[404.08, 264.55], [490.02, 264.23], [453.7, 297.94], [335.73, 298.54], [404.08, 180.01], [490.02, 179.98], [453.7, 182.61], [335.73, 182.66]]}, {\"category\": \"car\", \"corners_3d\": [[786.75, 289.02], [894.88, 288.68], [1028.73, 340.51], [872.58, 341.21], [786.75, 179.32], [894.88, 179.3], [1028.73, 182.19], [872.58, 182.23]]}, {\"category\": \"car\", \"corners_3d\": [[713.05, 253.26], [789.23, 253.1], [849.18, 277.96], [749.5, 278.24], [713.05, 179.14], [789.23, 179.12], [849.18, 181.06], [749.5, 181.09]]}, {\"category\": \"truck\", \"corners_3d\": [[522.53, 228.87], [446.41, 228.93], [480.13, 216.76], [539.71, 216.72], [522.53, 133.03], [446.41, 132.98], [480.13, 141.64], [539.71, 141.67]]}, {\"category\": \"car\", \"corners_3d\": [[661.26, 215.95], [702.21, 216.02], [712.86, 222.55], [665.71, 222.45], [661.26, 178.71], [702.21, 178.72], [712.86, 179.61], [665.71, 179.59]]}]\n```" - } - ] - }, - { - "id": 17, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[404.08, 264.55], [490.02, 264.23], [453.7, 297.94], [335.73, 298.54], [404.08, 180.01], [490.02, 179.98], [453.7, 182.61], [335.73, 182.66]]}, {\"category\": \"car\", \"corners_3d\": [[786.75, 289.02], [894.88, 288.68], [1028.73, 340.51], [872.58, 341.21], [786.75, 179.32], [894.88, 179.3], [1028.73, 182.19], [872.58, 182.23]]}, {\"category\": \"car\", \"corners_3d\": [[713.05, 253.26], [789.23, 253.1], [849.18, 277.96], [749.5, 278.24], [713.05, 179.14], [789.23, 179.12], [849.18, 181.06], [749.5, 181.09]]}, {\"category\": \"car\", \"corners_3d\": [[661.26, 215.95], [702.21, 216.02], [712.86, 222.55], [665.71, 222.45], [661.26, 178.71], [702.21, 178.72], [712.86, 179.61], [665.71, 179.59]]}, {\"category\": \"van\", \"corners_3d\": [[621.29, 198.32], [645.55, 198.32], [649.33, 200.57], [622.92, 200.58], [621.29, 172.05], [645.55, 172.05], [649.33, 171.98], [622.92, 171.98]]}]\n```" - } - ] - }, - { - "id": 18, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[522.53, 228.87], [446.41, 228.93], [480.13, 216.76], [539.71, 216.72], [522.53, 133.03], [446.41, 132.98], [480.13, 141.64], [539.71, 141.67]]}, {\"category\": \"van\", \"corners_3d\": [[621.29, 198.32], [645.55, 198.32], [649.33, 200.57], [622.92, 200.58], [621.29, 172.05], [645.55, 172.05], [649.33, 171.98], [622.92, 171.98]]}]\n```" - } - ] - }, - { - "id": 19, - "data_source": "Mono3DRefer", - "scene_name": "000031", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[404.08, 264.55], [490.02, 264.23], [453.7, 297.94], [335.73, 298.54], [404.08, 180.01], [490.02, 179.98], [453.7, 182.61], [335.73, 182.66]]}, {\"category\": \"car\", \"corners_3d\": [[786.75, 289.02], [894.88, 288.68], [1028.73, 340.51], [872.58, 341.21], [786.75, 179.32], [894.88, 179.3], [1028.73, 182.19], [872.58, 182.23]]}, {\"category\": \"car\", \"corners_3d\": [[713.05, 253.26], [789.23, 253.1], [849.18, 277.96], [749.5, 278.24], [713.05, 179.14], [789.23, 179.12], [849.18, 181.06], [749.5, 181.09]]}, {\"category\": \"truck\", \"corners_3d\": [[522.53, 228.87], [446.41, 228.93], [480.13, 216.76], [539.71, 216.72], [522.53, 133.03], [446.41, 132.98], [480.13, 141.64], [539.71, 141.67]]}, {\"category\": \"car\", \"corners_3d\": [[661.26, 215.95], [702.21, 216.02], [712.86, 222.55], [665.71, 222.45], [661.26, 178.71], [702.21, 178.72], [712.86, 179.61], [665.71, 179.59]]}, {\"category\": \"van\", \"corners_3d\": [[621.29, 198.32], [645.55, 198.32], [649.33, 200.57], [622.92, 200.58], [621.29, 172.05], [645.55, 172.05], [649.33, 171.98], [622.92, 171.98]]}]\n```" - } - ] - }, - { - "id": 20, - "data_source": "Mono3DRefer", - "scene_name": "000050", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[703.85, 182.86], [724.02, 182.99], [705.49, 183.76], [684.07, 183.61], [703.85, 155.57], [724.02, 155.34], [705.49, 154.01], [684.07, 154.27]]}]\n```" - } - ] - }, - { - "id": 21, - "data_source": "Mono3DRefer", - "scene_name": "000050", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[684.17, 236.43], [750.48, 236.31], [803.79, 258.17], [714.71, 258.37], [684.17, 172.84], [750.48, 172.84], [803.79, 172.83], [714.71, 172.83]]}, {\"category\": \"car\", \"corners_3d\": [[374.85, 268.05], [468.88, 267.57], [409.3, 317.79], [264.56, 318.92], [374.85, 181.38], [468.88, 181.33], [409.3, 185.83], [264.56, 185.93]]}, {\"category\": \"car\", \"corners_3d\": [[913.72, 432.06], [1173.49, 429.3], [5994.92, 2542.82], [3886.76, 2801.5], [913.72, 166.1], [1173.49, 166.18], [5994.92, 111.13], [3886.76, 104.4]]}, {\"category\": \"car\", \"corners_3d\": [[643.31, 202.98], [678.74, 203.07], [681.7, 206.61], [642.12, 206.49], [643.31, 173.28], [678.74, 173.28], [681.7, 173.33], [642.12, 173.33]]}]\n```" - } - ] - }, - { - "id": 22, - "data_source": "Mono3DRefer", - "scene_name": "000050", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[684.17, 236.43], [750.48, 236.31], [803.79, 258.17], [714.71, 258.37], [684.17, 172.84], [750.48, 172.84], [803.79, 172.83], [714.71, 172.83]]}, {\"category\": \"car\", \"corners_3d\": [[374.85, 268.05], [468.88, 267.57], [409.3, 317.79], [264.56, 318.92], [374.85, 181.38], [468.88, 181.33], [409.3, 185.83], [264.56, 185.93]]}, {\"category\": \"car\", \"corners_3d\": [[913.72, 432.06], [1173.49, 429.3], [5994.92, 2542.82], [3886.76, 2801.5], [913.72, 166.1], [1173.49, 166.18], [5994.92, 111.13], [3886.76, 104.4]]}, {\"category\": \"car\", \"corners_3d\": [[643.31, 202.98], [678.74, 203.07], [681.7, 206.61], [642.12, 206.49], [643.31, 173.28], [678.74, 173.28], [681.7, 173.33], [642.12, 173.33]]}, {\"category\": \"van\", \"corners_3d\": [[703.85, 182.86], [724.02, 182.99], [705.49, 183.76], [684.07, 183.61], [703.85, 155.57], [724.02, 155.34], [705.49, 154.01], [684.07, 154.27]]}]\n```" - } - ] - }, - { - "id": 23, - "data_source": "Mono3DRefer", - "scene_name": "000126", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[524.92, 195.24], [554.0, 194.75], [594.87, 197.19], [563.49, 197.8], [524.92, 152.04], [554.0, 152.5], [594.87, 150.23], [563.49, 149.66]]}]\n```" - } - ] - }, - { - "id": 24, - "data_source": "Mono3DRefer", - "scene_name": "000126", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[408.12, 266.98], [316.73, 268.15], [372.91, 243.8], [440.64, 243.15], [408.12, 194.89], [316.73, 195.17], [372.91, 189.47], [440.64, 189.31]]}]\n```" - } - ] - }, - { - "id": 25, - "data_source": "Mono3DRefer", - "scene_name": "000126", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[408.12, 266.98], [316.73, 268.15], [372.91, 243.8], [440.64, 243.15], [408.12, 194.89], [316.73, 195.17], [372.91, 189.47], [440.64, 189.31]]}, {\"category\": \"van\", \"corners_3d\": [[524.92, 195.24], [554.0, 194.75], [594.87, 197.19], [563.49, 197.8], [524.92, 152.04], [554.0, 152.5], [594.87, 150.23], [563.49, 149.66]]}]\n```" - } - ] - }, - { - "id": 26, - "data_source": "Mono3DRefer", - "scene_name": "000140", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[708.09, 232.0], [771.93, 232.0], [819.48, 249.39], [736.87, 249.38], [708.09, 166.22], [771.93, 166.22], [819.48, 164.27], [736.87, 164.28]]}, {\"category\": \"car\", \"corners_3d\": [[512.97, 235.44], [454.18, 235.49], [477.07, 225.78], [526.74, 225.74], [512.97, 180.0], [454.18, 180.0], [477.07, 178.89], [526.74, 178.89]]}, {\"category\": \"car\", \"corners_3d\": [[566.85, 207.25], [603.74, 207.41], [593.15, 212.52], [550.89, 212.31], [566.85, 173.87], [603.74, 173.88], [593.15, 174.03], [550.89, 174.02]]}, {\"category\": \"car\", \"corners_3d\": [[708.35, 205.51], [744.4, 206.2], [725.27, 209.59], [686.07, 208.75], [708.35, 170.72], [744.4, 170.67], [725.27, 170.45], [686.07, 170.5]]}]\n```" - } - ] - }, - { - "id": 27, - "data_source": "Mono3DRefer", - "scene_name": "000152", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[315.43, 323.89], [448.52, 323.08], [311.27, 470.4], [44.81, 473.57], [315.43, 195.87], [448.52, 195.74], [311.27, 218.19], [44.81, 218.68]]}, {\"category\": \"car\", \"corners_3d\": [[753.16, 326.3], [898.16, 326.02], [1137.58, 450.22], [875.41, 451.16], [753.16, 165.77], [898.16, 165.78], [1137.58, 160.04], [875.41, 160.0]]}, {\"category\": \"car\", \"corners_3d\": [[712.26, 265.76], [791.8, 265.67], [877.8, 307.93], [762.12, 308.13], [712.26, 185.51], [791.8, 185.49], [877.8, 191.25], [762.12, 191.28]]}, {\"category\": \"car\", \"corners_3d\": [[517.26, 244.27], [454.46, 244.34], [481.71, 231.11], [532.88, 231.07], [517.26, 187.6], [454.46, 187.62], [481.71, 184.89], [532.88, 184.88]]}, {\"category\": \"car\", \"corners_3d\": [[663.4, 238.75], [726.12, 238.35], [773.25, 256.88], [693.02, 257.54], [663.4, 183.31], [726.12, 183.24], [773.25, 186.18], [693.02, 186.29]]}, {\"category\": \"car\", \"corners_3d\": [[561.84, 204.63], [533.8, 204.68], [537.65, 202.03], [563.36, 201.99], [561.84, 179.42], [533.8, 179.43], [537.65, 178.89], [563.36, 178.88]]}, {\"category\": \"car\", \"corners_3d\": [[628.85, 198.7], [653.21, 198.69], [657.77, 201.03], [631.21, 201.04], [628.85, 176.69], [653.21, 176.69], [657.77, 177.04], [631.21, 177.04]]}]\n```" - } - ] - }, - { - "id": 28, - "data_source": "Mono3DRefer", - "scene_name": "000159", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[749.75, 239.38], [822.55, 239.71], [893.81, 266.74], [791.3, 266.07], [749.75, 167.57], [822.55, 167.54], [893.81, 165.39], [791.3, 165.44]]}, {\"category\": \"car\", \"corners_3d\": [[-44.22, 478.09], [-305.01, 475.93], [125.92, 335.26], [266.85, 335.87], [-44.22, 238.34], [-305.01, 237.87], [125.92, 207.69], [266.85, 207.83]]}, {\"category\": \"car\", \"corners_3d\": [[303.77, 315.64], [170.89, 314.86], [313.23, 270.9], [405.35, 271.27], [303.77, 194.41], [170.89, 194.3], [313.23, 187.66], [405.35, 187.71]]}, {\"category\": \"car\", \"corners_3d\": [[560.1, 203.69], [526.6, 203.58], [540.01, 200.64], [570.33, 200.73], [560.1, 173.66], [526.6, 173.66], [540.01, 173.58], [570.33, 173.59]]}, {\"category\": \"car\", \"corners_3d\": [[725.5, 220.16], [777.47, 220.37], [802.12, 229.82], [739.74, 229.52], [725.5, 176.55], [777.47, 176.57], [802.12, 177.31], [739.74, 177.29]]}, {\"category\": \"car\", \"corners_3d\": [[709.79, 211.6], [749.89, 211.62], [767.38, 216.74], [721.98, 216.71], [709.79, 171.59], [749.89, 171.59], [767.38, 171.42], [721.98, 171.42]]}, {\"category\": \"car\", \"corners_3d\": [[702.75, 202.68], [736.9, 202.77], [747.07, 206.41], [708.73, 206.31], [702.75, 171.58], [736.9, 171.57], [747.07, 171.42], [708.73, 171.42]]}, {\"category\": \"car\", \"corners_3d\": [[610.3, 187.25], [592.04, 187.21], [597.05, 186.52], [614.43, 186.56], [610.3, 170.55], [592.04, 170.55], [597.05, 170.66], [614.43, 170.66]]}]\n```" - } - ] - }, - { - "id": 29, - "data_source": "Mono3DRefer", - "scene_name": "000175", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[477.76, 218.82], [511.61, 218.67], [506.33, 223.71], [468.72, 223.9], [477.76, 184.77], [511.61, 184.73], [506.33, 186.04], [468.72, 186.09]]}, {\"category\": \"car\", \"corners_3d\": [[515.96, 214.68], [482.44, 214.76], [491.92, 210.69], [522.18, 210.63], [515.96, 183.75], [482.44, 183.77], [491.92, 182.71], [522.18, 182.69]]}, {\"category\": \"car\", \"corners_3d\": [[265.04, 229.15], [277.48, 225.46], [408.83, 223.58], [405.74, 227.01], [265.04, 176.14], [277.48, 175.92], [408.83, 175.81], [405.74, 176.01]]}]\n```" - } - ] - }, - { - "id": 30, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-71.34, 415.53], [-297.89, 416.34], [73.09, 315.94], [205.67, 315.66], [-71.34, 211.2], [-297.89, 211.33], [73.09, 195.46], [205.67, 195.42]]}, {\"category\": \"car\", \"corners_3d\": [[891.42, 383.75], [1093.62, 380.17], [2000.58, 737.06], [1477.39, 764.44], [891.42, 188.53], [1093.62, 188.26], [2000.58, 214.79], [1477.39, 216.83]]}, {\"category\": \"car\", \"corners_3d\": [[514.01, 216.06], [477.42, 216.94], [460.51, 212.64], [493.9, 211.92], [514.01, 182.94], [477.42, 183.15], [460.51, 182.15], [493.9, 181.98]]}, {\"category\": \"car\", \"corners_3d\": [[545.97, 199.68], [534.45, 198.89], [587.7, 197.76], [600.25, 198.48], [545.97, 177.52], [534.45, 177.38], [587.7, 177.19], [600.25, 177.31]]}, {\"category\": \"car\", \"corners_3d\": [[418.79, 210.21], [396.63, 211.74], [337.91, 209.96], [361.31, 208.56], [418.79, 183.94], [396.63, 184.4], [337.91, 183.87], [361.31, 183.46]]}, {\"category\": \"car\", \"corners_3d\": [[169.93, 233.95], [101.79, 235.18], [147.9, 226.49], [205.91, 225.57], [169.93, 182.44], [101.79, 182.63], [147.9, 181.27], [205.91, 181.13]]}, {\"category\": \"car\", \"corners_3d\": [[651.0, 195.92], [641.39, 195.16], [699.39, 194.53], [710.65, 195.25], [651.0, 175.23], [641.39, 175.15], [699.39, 175.08], [710.65, 175.16]]}]\n```" - } - ] - }, - { - "id": 31, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[511.73, 199.51], [515.56, 198.46], [597.79, 198.45], [597.35, 199.51], [511.73, 155.73], [515.56, 156.4], [597.79, 156.41], [597.35, 155.73]]}]\n```" - } - ] - }, - { - "id": 32, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[699.95, 211.18], [685.83, 211.59], [674.96, 210.64], [688.84, 210.25], [699.95, 170.18], [685.83, 170.15], [674.96, 170.21], [688.84, 170.24]]}]\n```" - } - ] - }, - { - "id": 33, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-71.34, 415.53], [-297.89, 416.34], [73.09, 315.94], [205.67, 315.66], [-71.34, 211.2], [-297.89, 211.33], [73.09, 195.46], [205.67, 195.42]]}, {\"category\": \"car\", \"corners_3d\": [[891.42, 383.75], [1093.62, 380.17], [2000.58, 737.06], [1477.39, 764.44], [891.42, 188.53], [1093.62, 188.26], [2000.58, 214.79], [1477.39, 216.83]]}, {\"category\": \"car\", \"corners_3d\": [[514.01, 216.06], [477.42, 216.94], [460.51, 212.64], [493.9, 211.92], [514.01, 182.94], [477.42, 183.15], [460.51, 182.15], [493.9, 181.98]]}, {\"category\": \"car\", \"corners_3d\": [[545.97, 199.68], [534.45, 198.89], [587.7, 197.76], [600.25, 198.48], [545.97, 177.52], [534.45, 177.38], [587.7, 177.19], [600.25, 177.31]]}, {\"category\": \"car\", \"corners_3d\": [[418.79, 210.21], [396.63, 211.74], [337.91, 209.96], [361.31, 208.56], [418.79, 183.94], [396.63, 184.4], [337.91, 183.87], [361.31, 183.46]]}, {\"category\": \"car\", \"corners_3d\": [[169.93, 233.95], [101.79, 235.18], [147.9, 226.49], [205.91, 225.57], [169.93, 182.44], [101.79, 182.63], [147.9, 181.27], [205.91, 181.13]]}, {\"category\": \"car\", \"corners_3d\": [[651.0, 195.92], [641.39, 195.16], [699.39, 194.53], [710.65, 195.25], [651.0, 175.23], [641.39, 175.15], [699.39, 175.08], [710.65, 175.16]]}, {\"category\": \"truck\", \"corners_3d\": [[511.73, 199.51], [515.56, 198.46], [597.79, 198.45], [597.35, 199.51], [511.73, 155.73], [515.56, 156.4], [597.79, 156.41], [597.35, 155.73]]}]\n```" - } - ] - }, - { - "id": 34, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-71.34, 415.53], [-297.89, 416.34], [73.09, 315.94], [205.67, 315.66], [-71.34, 211.2], [-297.89, 211.33], [73.09, 195.46], [205.67, 195.42]]}, {\"category\": \"car\", \"corners_3d\": [[891.42, 383.75], [1093.62, 380.17], [2000.58, 737.06], [1477.39, 764.44], [891.42, 188.53], [1093.62, 188.26], [2000.58, 214.79], [1477.39, 216.83]]}, {\"category\": \"car\", \"corners_3d\": [[514.01, 216.06], [477.42, 216.94], [460.51, 212.64], [493.9, 211.92], [514.01, 182.94], [477.42, 183.15], [460.51, 182.15], [493.9, 181.98]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[699.95, 211.18], [685.83, 211.59], [674.96, 210.64], [688.84, 210.25], [699.95, 170.18], [685.83, 170.15], [674.96, 170.21], [688.84, 170.24]]}, {\"category\": \"car\", \"corners_3d\": [[545.97, 199.68], [534.45, 198.89], [587.7, 197.76], [600.25, 198.48], [545.97, 177.52], [534.45, 177.38], [587.7, 177.19], [600.25, 177.31]]}, {\"category\": \"car\", \"corners_3d\": [[418.79, 210.21], [396.63, 211.74], [337.91, 209.96], [361.31, 208.56], [418.79, 183.94], [396.63, 184.4], [337.91, 183.87], [361.31, 183.46]]}, {\"category\": \"car\", \"corners_3d\": [[169.93, 233.95], [101.79, 235.18], [147.9, 226.49], [205.91, 225.57], [169.93, 182.44], [101.79, 182.63], [147.9, 181.27], [205.91, 181.13]]}, {\"category\": \"car\", \"corners_3d\": [[651.0, 195.92], [641.39, 195.16], [699.39, 194.53], [710.65, 195.25], [651.0, 175.23], [641.39, 175.15], [699.39, 175.08], [710.65, 175.16]]}]\n```" - } - ] - }, - { - "id": 35, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[699.95, 211.18], [685.83, 211.59], [674.96, 210.64], [688.84, 210.25], [699.95, 170.18], [685.83, 170.15], [674.96, 170.21], [688.84, 170.24]]}, {\"category\": \"truck\", \"corners_3d\": [[511.73, 199.51], [515.56, 198.46], [597.79, 198.45], [597.35, 199.51], [511.73, 155.73], [515.56, 156.4], [597.79, 156.41], [597.35, 155.73]]}]\n```" - } - ] - }, - { - "id": 36, - "data_source": "Mono3DRefer", - "scene_name": "000186", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-71.34, 415.53], [-297.89, 416.34], [73.09, 315.94], [205.67, 315.66], [-71.34, 211.2], [-297.89, 211.33], [73.09, 195.46], [205.67, 195.42]]}, {\"category\": \"car\", \"corners_3d\": [[891.42, 383.75], [1093.62, 380.17], [2000.58, 737.06], [1477.39, 764.44], [891.42, 188.53], [1093.62, 188.26], [2000.58, 214.79], [1477.39, 216.83]]}, {\"category\": \"car\", \"corners_3d\": [[514.01, 216.06], [477.42, 216.94], [460.51, 212.64], [493.9, 211.92], [514.01, 182.94], [477.42, 183.15], [460.51, 182.15], [493.9, 181.98]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[699.95, 211.18], [685.83, 211.59], [674.96, 210.64], [688.84, 210.25], [699.95, 170.18], [685.83, 170.15], [674.96, 170.21], [688.84, 170.24]]}, {\"category\": \"car\", \"corners_3d\": [[545.97, 199.68], [534.45, 198.89], [587.7, 197.76], [600.25, 198.48], [545.97, 177.52], [534.45, 177.38], [587.7, 177.19], [600.25, 177.31]]}, {\"category\": \"car\", \"corners_3d\": [[418.79, 210.21], [396.63, 211.74], [337.91, 209.96], [361.31, 208.56], [418.79, 183.94], [396.63, 184.4], [337.91, 183.87], [361.31, 183.46]]}, {\"category\": \"car\", \"corners_3d\": [[169.93, 233.95], [101.79, 235.18], [147.9, 226.49], [205.91, 225.57], [169.93, 182.44], [101.79, 182.63], [147.9, 181.27], [205.91, 181.13]]}, {\"category\": \"car\", \"corners_3d\": [[651.0, 195.92], [641.39, 195.16], [699.39, 194.53], [710.65, 195.25], [651.0, 175.23], [641.39, 175.15], [699.39, 175.08], [710.65, 175.16]]}, {\"category\": \"truck\", \"corners_3d\": [[511.73, 199.51], [515.56, 198.46], [597.79, 198.45], [597.35, 199.51], [511.73, 155.73], [515.56, 156.4], [597.79, 156.41], [597.35, 155.73]]}]\n```" - } - ] - }, - { - "id": 37, - "data_source": "Mono3DRefer", - "scene_name": "000187", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[258.19, 261.16], [216.25, 261.06], [231.8, 257.69], [272.16, 257.78], [258.19, 178.58], [216.25, 178.57], [231.8, 178.35], [272.16, 178.36]]}]\n```" - } - ] - }, - { - "id": 38, - "data_source": "Mono3DRefer", - "scene_name": "000188", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[847.85, 248.7], [933.97, 249.75], [1037.28, 283.3], [912.3, 281.14], [847.85, 163.97], [933.97, 163.85], [1037.28, 159.92], [912.3, 160.17]]}, {\"category\": \"car\", \"corners_3d\": [[651.25, 247.23], [582.68, 246.32], [605.41, 232.47], [661.16, 233.07], [651.25, 183.08], [582.68, 182.96], [605.41, 181.05], [661.16, 181.14]]}, {\"category\": \"car\", \"corners_3d\": [[660.56, 205.28], [691.03, 205.41], [692.12, 209.6], [657.75, 209.42], [660.56, 175.97], [691.03, 175.99], [692.12, 176.39], [657.75, 176.37]]}, {\"category\": \"car\", \"corners_3d\": [[769.22, 206.59], [803.4, 206.86], [811.23, 210.7], [773.17, 210.37], [769.22, 170.03], [803.4, 170.01], [811.23, 169.69], [773.17, 169.72]]}]\n```" - } - ] - }, - { - "id": 39, - "data_source": "Mono3DRefer", - "scene_name": "000196", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[580.93, 237.14], [544.84, 237.2], [550.53, 229.84], [582.49, 229.78], [580.93, 203.77], [544.84, 203.8], [550.53, 200.26], [582.49, 200.23]]}]\n```" - } - ] - }, - { - "id": 40, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[415.07, 208.85], [448.26, 208.44], [450.95, 211.62], [414.78, 212.11], [415.07, 183.95], [448.26, 183.83], [450.95, 184.81], [414.78, 184.96]]}, {\"category\": \"car\", \"corners_3d\": [[537.29, 203.17], [561.83, 202.96], [570.06, 205.31], [543.66, 205.56], [537.29, 179.71], [561.83, 179.66], [570.06, 180.19], [543.66, 180.25]]}, {\"category\": \"car\", \"corners_3d\": [[1331.43, 470.01], [1041.72, 475.08], [807.29, 321.6], [953.06, 320.36], [1331.43, 180.36], [1041.72, 180.49], [807.29, 176.61], [953.06, 176.58]]}, {\"category\": \"car\", \"corners_3d\": [[660.95, 245.15], [729.4, 244.11], [786.09, 261.8], [701.37, 263.43], [660.95, 181.22], [729.4, 181.1], [786.09, 183.14], [701.37, 183.33]]}, {\"category\": \"car\", \"corners_3d\": [[348.67, 242.49], [414.22, 241.53], [396.43, 256.9], [315.65, 258.34], [348.67, 192.04], [414.22, 191.78], [396.43, 196.01], [315.65, 196.41]]}, {\"category\": \"car\", \"corners_3d\": [[609.7, 223.02], [656.87, 222.39], [690.91, 231.48], [635.46, 232.36], [609.7, 171.31], [656.87, 171.33], [690.91, 171.05], [635.46, 171.03]]}, {\"category\": \"car\", \"corners_3d\": [[560.58, 210.84], [594.9, 210.52], [607.55, 214.63], [569.58, 215.03], [560.58, 180.57], [594.9, 180.5], [607.55, 181.34], [569.58, 181.42]]}]\n```" - } - ] - }, - { - "id": 41, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[417.48, 201.13], [406.81, 201.22], [407.41, 200.81], [417.92, 200.72], [417.48, 173.91], [406.81, 173.91], [407.41, 173.9], [417.92, 173.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[408.43, 205.62], [398.23, 205.69], [399.53, 205.24], [409.58, 205.17], [408.43, 178.31], [398.23, 178.32], [399.53, 178.25], [409.58, 178.24]]}]\n```" - } - ] - }, - { - "id": 42, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[473.26, 238.57], [448.47, 238.79], [456.64, 232.68], [479.12, 232.49], [473.26, 176.16], [448.47, 176.18], [456.64, 175.87], [479.12, 175.86]]}, {\"category\": \"cyclist\", \"corners_3d\": [[437.44, 249.37], [413.51, 249.59], [425.82, 242.35], [447.46, 242.16], [437.44, 176.95], [413.51, 176.97], [425.82, 176.58], [447.46, 176.57]]}]\n```" - } - ] - }, - { - "id": 43, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[415.07, 208.85], [448.26, 208.44], [450.95, 211.62], [414.78, 212.11], [415.07, 183.95], [448.26, 183.83], [450.95, 184.81], [414.78, 184.96]]}, {\"category\": \"car\", \"corners_3d\": [[537.29, 203.17], [561.83, 202.96], [570.06, 205.31], [543.66, 205.56], [537.29, 179.71], [561.83, 179.66], [570.06, 180.19], [543.66, 180.25]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[417.48, 201.13], [406.81, 201.22], [407.41, 200.81], [417.92, 200.72], [417.48, 173.91], [406.81, 173.91], [407.41, 173.9], [417.92, 173.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[408.43, 205.62], [398.23, 205.69], [399.53, 205.24], [409.58, 205.17], [408.43, 178.31], [398.23, 178.32], [399.53, 178.25], [409.58, 178.24]]}, {\"category\": \"car\", \"corners_3d\": [[1331.43, 470.01], [1041.72, 475.08], [807.29, 321.6], [953.06, 320.36], [1331.43, 180.36], [1041.72, 180.49], [807.29, 176.61], [953.06, 176.58]]}, {\"category\": \"car\", \"corners_3d\": [[660.95, 245.15], [729.4, 244.11], [786.09, 261.8], [701.37, 263.43], [660.95, 181.22], [729.4, 181.1], [786.09, 183.14], [701.37, 183.33]]}, {\"category\": \"car\", \"corners_3d\": [[348.67, 242.49], [414.22, 241.53], [396.43, 256.9], [315.65, 258.34], [348.67, 192.04], [414.22, 191.78], [396.43, 196.01], [315.65, 196.41]]}, {\"category\": \"car\", \"corners_3d\": [[609.7, 223.02], [656.87, 222.39], [690.91, 231.48], [635.46, 232.36], [609.7, 171.31], [656.87, 171.33], [690.91, 171.05], [635.46, 171.03]]}, {\"category\": \"car\", \"corners_3d\": [[560.58, 210.84], [594.9, 210.52], [607.55, 214.63], [569.58, 215.03], [560.58, 180.57], [594.9, 180.5], [607.55, 181.34], [569.58, 181.42]]}]\n```" - } - ] - }, - { - "id": 44, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[415.07, 208.85], [448.26, 208.44], [450.95, 211.62], [414.78, 212.11], [415.07, 183.95], [448.26, 183.83], [450.95, 184.81], [414.78, 184.96]]}, {\"category\": \"car\", \"corners_3d\": [[537.29, 203.17], [561.83, 202.96], [570.06, 205.31], [543.66, 205.56], [537.29, 179.71], [561.83, 179.66], [570.06, 180.19], [543.66, 180.25]]}, {\"category\": \"cyclist\", \"corners_3d\": [[473.26, 238.57], [448.47, 238.79], [456.64, 232.68], [479.12, 232.49], [473.26, 176.16], [448.47, 176.18], [456.64, 175.87], [479.12, 175.86]]}, {\"category\": \"car\", \"corners_3d\": [[1331.43, 470.01], [1041.72, 475.08], [807.29, 321.6], [953.06, 320.36], [1331.43, 180.36], [1041.72, 180.49], [807.29, 176.61], [953.06, 176.58]]}, {\"category\": \"car\", \"corners_3d\": [[660.95, 245.15], [729.4, 244.11], [786.09, 261.8], [701.37, 263.43], [660.95, 181.22], [729.4, 181.1], [786.09, 183.14], [701.37, 183.33]]}, {\"category\": \"cyclist\", \"corners_3d\": [[437.44, 249.37], [413.51, 249.59], [425.82, 242.35], [447.46, 242.16], [437.44, 176.95], [413.51, 176.97], [425.82, 176.58], [447.46, 176.57]]}, {\"category\": \"car\", \"corners_3d\": [[348.67, 242.49], [414.22, 241.53], [396.43, 256.9], [315.65, 258.34], [348.67, 192.04], [414.22, 191.78], [396.43, 196.01], [315.65, 196.41]]}, {\"category\": \"car\", \"corners_3d\": [[609.7, 223.02], [656.87, 222.39], [690.91, 231.48], [635.46, 232.36], [609.7, 171.31], [656.87, 171.33], [690.91, 171.05], [635.46, 171.03]]}, {\"category\": \"car\", \"corners_3d\": [[560.58, 210.84], [594.9, 210.52], [607.55, 214.63], [569.58, 215.03], [560.58, 180.57], [594.9, 180.5], [607.55, 181.34], [569.58, 181.42]]}]\n```" - } - ] - }, - { - "id": 45, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[417.48, 201.13], [406.81, 201.22], [407.41, 200.81], [417.92, 200.72], [417.48, 173.91], [406.81, 173.91], [407.41, 173.9], [417.92, 173.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[408.43, 205.62], [398.23, 205.69], [399.53, 205.24], [409.58, 205.17], [408.43, 178.31], [398.23, 178.32], [399.53, 178.25], [409.58, 178.24]]}, {\"category\": \"cyclist\", \"corners_3d\": [[473.26, 238.57], [448.47, 238.79], [456.64, 232.68], [479.12, 232.49], [473.26, 176.16], [448.47, 176.18], [456.64, 175.87], [479.12, 175.86]]}, {\"category\": \"cyclist\", \"corners_3d\": [[437.44, 249.37], [413.51, 249.59], [425.82, 242.35], [447.46, 242.16], [437.44, 176.95], [413.51, 176.97], [425.82, 176.58], [447.46, 176.57]]}]\n```" - } - ] - }, - { - "id": 46, - "data_source": "Mono3DRefer", - "scene_name": "000201", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[415.07, 208.85], [448.26, 208.44], [450.95, 211.62], [414.78, 212.11], [415.07, 183.95], [448.26, 183.83], [450.95, 184.81], [414.78, 184.96]]}, {\"category\": \"car\", \"corners_3d\": [[537.29, 203.17], [561.83, 202.96], [570.06, 205.31], [543.66, 205.56], [537.29, 179.71], [561.83, 179.66], [570.06, 180.19], [543.66, 180.25]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[417.48, 201.13], [406.81, 201.22], [407.41, 200.81], [417.92, 200.72], [417.48, 173.91], [406.81, 173.91], [407.41, 173.9], [417.92, 173.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[408.43, 205.62], [398.23, 205.69], [399.53, 205.24], [409.58, 205.17], [408.43, 178.31], [398.23, 178.32], [399.53, 178.25], [409.58, 178.24]]}, {\"category\": \"cyclist\", \"corners_3d\": [[473.26, 238.57], [448.47, 238.79], [456.64, 232.68], [479.12, 232.49], [473.26, 176.16], [448.47, 176.18], [456.64, 175.87], [479.12, 175.86]]}, {\"category\": \"car\", \"corners_3d\": [[1331.43, 470.01], [1041.72, 475.08], [807.29, 321.6], [953.06, 320.36], [1331.43, 180.36], [1041.72, 180.49], [807.29, 176.61], [953.06, 176.58]]}, {\"category\": \"car\", \"corners_3d\": [[660.95, 245.15], [729.4, 244.11], [786.09, 261.8], [701.37, 263.43], [660.95, 181.22], [729.4, 181.1], [786.09, 183.14], [701.37, 183.33]]}, {\"category\": \"cyclist\", \"corners_3d\": [[437.44, 249.37], [413.51, 249.59], [425.82, 242.35], [447.46, 242.16], [437.44, 176.95], [413.51, 176.97], [425.82, 176.58], [447.46, 176.57]]}, {\"category\": \"car\", \"corners_3d\": [[348.67, 242.49], [414.22, 241.53], [396.43, 256.9], [315.65, 258.34], [348.67, 192.04], [414.22, 191.78], [396.43, 196.01], [315.65, 196.41]]}, {\"category\": \"car\", \"corners_3d\": [[609.7, 223.02], [656.87, 222.39], [690.91, 231.48], [635.46, 232.36], [609.7, 171.31], [656.87, 171.33], [690.91, 171.05], [635.46, 171.03]]}, {\"category\": \"car\", \"corners_3d\": [[560.58, 210.84], [594.9, 210.52], [607.55, 214.63], [569.58, 215.03], [560.58, 180.57], [594.9, 180.5], [607.55, 181.34], [569.58, 181.42]]}]\n```" - } - ] - }, - { - "id": 47, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[766.45, 243.47], [789.81, 243.45], [812.73, 252.11], [786.51, 252.13], [766.45, 170.89], [789.81, 170.89], [812.73, 170.65], [786.51, 170.65]]}]\n```" - } - ] - }, - { - "id": 48, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[588.56, 202.54], [615.85, 202.51], [620.19, 208.91], [587.02, 208.94], [588.56, 172.95], [615.85, 172.95], [620.19, 172.98], [587.02, 172.98]]}]\n```" - } - ] - }, - { - "id": 49, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[386.77, 217.85], [362.23, 217.85], [379.11, 214.77], [401.97, 214.77], [386.77, 195.28], [362.23, 195.28], [379.11, 193.74], [401.97, 193.74]]}]\n```" - } - ] - }, - { - "id": 50, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[588.56, 202.54], [615.85, 202.51], [620.19, 208.91], [587.02, 208.94], [588.56, 172.95], [615.85, 172.95], [620.19, 172.98], [587.02, 172.98]]}, {\"category\": \"cyclist\", \"corners_3d\": [[766.45, 243.47], [789.81, 243.45], [812.73, 252.11], [786.51, 252.13], [766.45, 170.89], [789.81, 170.89], [812.73, 170.65], [786.51, 170.65]]}]\n```" - } - ] - }, - { - "id": 51, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[386.77, 217.85], [362.23, 217.85], [379.11, 214.77], [401.97, 214.77], [386.77, 195.28], [362.23, 195.28], [379.11, 193.74], [401.97, 193.74]]}, {\"category\": \"cyclist\", \"corners_3d\": [[766.45, 243.47], [789.81, 243.45], [812.73, 252.11], [786.51, 252.13], [766.45, 170.89], [789.81, 170.89], [812.73, 170.65], [786.51, 170.65]]}]\n```" - } - ] - }, - { - "id": 52, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[386.77, 217.85], [362.23, 217.85], [379.11, 214.77], [401.97, 214.77], [386.77, 195.28], [362.23, 195.28], [379.11, 193.74], [401.97, 193.74]]}, {\"category\": \"bus\", \"corners_3d\": [[588.56, 202.54], [615.85, 202.51], [620.19, 208.91], [587.02, 208.94], [588.56, 172.95], [615.85, 172.95], [620.19, 172.98], [587.02, 172.98]]}]\n```" - } - ] - }, - { - "id": 53, - "data_source": "Mono3DRefer", - "scene_name": "000203", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[386.77, 217.85], [362.23, 217.85], [379.11, 214.77], [401.97, 214.77], [386.77, 195.28], [362.23, 195.28], [379.11, 193.74], [401.97, 193.74]]}, {\"category\": \"bus\", \"corners_3d\": [[588.56, 202.54], [615.85, 202.51], [620.19, 208.91], [587.02, 208.94], [588.56, 172.95], [615.85, 172.95], [620.19, 172.98], [587.02, 172.98]]}, {\"category\": \"cyclist\", \"corners_3d\": [[766.45, 243.47], [789.81, 243.45], [812.73, 252.11], [786.51, 252.13], [766.45, 170.89], [789.81, 170.89], [812.73, 170.65], [786.51, 170.65]]}]\n```" - } - ] - }, - { - "id": 54, - "data_source": "Mono3DRefer", - "scene_name": "000207", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[728.94, 210.2], [714.97, 210.22], [713.01, 209.61], [726.75, 209.59], [728.94, 177.7], [714.97, 177.71], [713.01, 177.63], [726.75, 177.62]]}]\n```" - } - ] - }, - { - "id": 55, - "data_source": "Mono3DRefer", - "scene_name": "000207", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[241.45, 234.53], [190.54, 234.58], [244.85, 226.42], [289.01, 226.39], [241.45, 189.05], [190.54, 189.06], [244.85, 186.92], [289.01, 186.91]]}, {\"category\": \"car\", \"corners_3d\": [[198.17, 231.12], [152.31, 231.12], [207.76, 224.05], [248.05, 224.05], [198.17, 188.38], [152.31, 188.39], [207.76, 186.5], [248.05, 186.5]]}, {\"category\": \"car\", \"corners_3d\": [[410.41, 208.34], [383.1, 208.31], [401.09, 205.66], [426.37, 205.68], [410.41, 184.48], [383.1, 184.47], [401.09, 183.6], [426.37, 183.61]]}]\n```" - } - ] - }, - { - "id": 56, - "data_source": "Mono3DRefer", - "scene_name": "000207", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[728.94, 210.2], [714.97, 210.22], [713.01, 209.61], [726.75, 209.59], [728.94, 177.7], [714.97, 177.71], [713.01, 177.63], [726.75, 177.62]]}, {\"category\": \"car\", \"corners_3d\": [[241.45, 234.53], [190.54, 234.58], [244.85, 226.42], [289.01, 226.39], [241.45, 189.05], [190.54, 189.06], [244.85, 186.92], [289.01, 186.91]]}, {\"category\": \"car\", \"corners_3d\": [[198.17, 231.12], [152.31, 231.12], [207.76, 224.05], [248.05, 224.05], [198.17, 188.38], [152.31, 188.39], [207.76, 186.5], [248.05, 186.5]]}, {\"category\": \"car\", \"corners_3d\": [[410.41, 208.34], [383.1, 208.31], [401.09, 205.66], [426.37, 205.68], [410.41, 184.48], [383.1, 184.47], [401.09, 183.6], [426.37, 183.61]]}]\n```" - } - ] - }, - { - "id": 57, - "data_source": "Mono3DRefer", - "scene_name": "000224", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[366.29, 203.17], [338.02, 203.19], [359.09, 200.7], [385.02, 200.68], [366.29, 181.44], [338.02, 181.45], [359.09, 180.74], [385.02, 180.74]]}]\n```" - } - ] - }, - { - "id": 58, - "data_source": "Mono3DRefer", - "scene_name": "000270", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[587.74, 196.14], [612.36, 196.13], [613.69, 198.0], [587.1, 198.02], [587.74, 171.82], [612.36, 171.82], [613.69, 171.73], [587.1, 171.73]]}, {\"category\": \"car\", \"corners_3d\": [[383.98, 224.38], [339.24, 224.45], [373.1, 217.61], [411.88, 217.56], [383.98, 184.91], [339.24, 184.92], [373.1, 183.32], [411.88, 183.31]]}, {\"category\": \"car\", \"corners_3d\": [[473.42, 201.61], [448.81, 201.69], [456.2, 199.61], [479.02, 199.54], [473.42, 177.21], [448.81, 177.22], [456.2, 176.9], [479.02, 176.89]]}]\n```" - } - ] - }, - { - "id": 59, - "data_source": "Mono3DRefer", - "scene_name": "000279", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[579.39, 221.39], [529.74, 221.32], [544.62, 213.62], [586.39, 213.67], [579.39, 176.81], [529.74, 176.81], [544.62, 176.18], [586.39, 176.18]]}, {\"category\": \"car\", \"corners_3d\": [[598.82, 202.5], [561.79, 202.29], [578.42, 199.0], [611.4, 199.17], [598.82, 167.25], [561.79, 167.29], [578.42, 167.91], [611.4, 167.88]]}, {\"category\": \"car\", \"corners_3d\": [[640.61, 191.08], [648.35, 191.65], [610.29, 191.95], [603.61, 191.36], [640.61, 168.0], [648.35, 167.85], [610.29, 167.77], [603.61, 167.92]]}, {\"category\": \"car\", \"corners_3d\": [[136.58, 262.98], [191.49, 251.99], [416.57, 251.22], [393.18, 261.99], [136.58, 175.22], [191.49, 174.93], [416.57, 174.91], [393.18, 175.19]]}]\n```" - } - ] - }, - { - "id": 60, - "data_source": "Mono3DRefer", - "scene_name": "000279", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[533.9, 248.98], [446.41, 248.71], [485.39, 232.66], [554.43, 232.83], [533.9, 160.39], [446.41, 160.43], [485.39, 163.06], [554.43, 163.03]]}, {\"category\": \"van\", \"corners_3d\": [[701.1, 239.17], [764.71, 239.17], [804.16, 256.1], [724.32, 256.09], [701.1, 170.11], [764.71, 170.11], [804.16, 169.41], [724.32, 169.41]]}, {\"category\": \"van\", \"corners_3d\": [[588.29, 210.56], [547.05, 210.5], [557.72, 205.69], [593.71, 205.74], [588.29, 165.55], [547.05, 165.56], [557.72, 166.49], [593.71, 166.48]]}]\n```" - } - ] - }, - { - "id": 61, - "data_source": "Mono3DRefer", - "scene_name": "000279", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[533.9, 248.98], [446.41, 248.71], [485.39, 232.66], [554.43, 232.83], [533.9, 160.39], [446.41, 160.43], [485.39, 163.06], [554.43, 163.03]]}, {\"category\": \"van\", \"corners_3d\": [[701.1, 239.17], [764.71, 239.17], [804.16, 256.1], [724.32, 256.09], [701.1, 170.11], [764.71, 170.11], [804.16, 169.41], [724.32, 169.41]]}, {\"category\": \"car\", \"corners_3d\": [[579.39, 221.39], [529.74, 221.32], [544.62, 213.62], [586.39, 213.67], [579.39, 176.81], [529.74, 176.81], [544.62, 176.18], [586.39, 176.18]]}, {\"category\": \"van\", \"corners_3d\": [[588.29, 210.56], [547.05, 210.5], [557.72, 205.69], [593.71, 205.74], [588.29, 165.55], [547.05, 165.56], [557.72, 166.49], [593.71, 166.48]]}, {\"category\": \"car\", \"corners_3d\": [[598.82, 202.5], [561.79, 202.29], [578.42, 199.0], [611.4, 199.17], [598.82, 167.25], [561.79, 167.29], [578.42, 167.91], [611.4, 167.88]]}, {\"category\": \"car\", \"corners_3d\": [[640.61, 191.08], [648.35, 191.65], [610.29, 191.95], [603.61, 191.36], [640.61, 168.0], [648.35, 167.85], [610.29, 167.77], [603.61, 167.92]]}, {\"category\": \"car\", \"corners_3d\": [[136.58, 262.98], [191.49, 251.99], [416.57, 251.22], [393.18, 261.99], [136.58, 175.22], [191.49, 174.93], [416.57, 174.91], [393.18, 175.19]]}]\n```" - } - ] - }, - { - "id": 62, - "data_source": "Mono3DRefer", - "scene_name": "000283", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[549.12, 192.02], [573.2, 191.67], [607.18, 193.42], [581.49, 193.83], [549.12, 156.06], [573.2, 156.36], [607.18, 154.83], [581.49, 154.47]]}]\n```" - } - ] - }, - { - "id": 63, - "data_source": "Mono3DRefer", - "scene_name": "000283", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[158.02, 264.69], [169.41, 257.48], [363.43, 249.81], [368.07, 255.72], [158.02, 192.72], [169.41, 191.16], [363.43, 189.5], [368.07, 190.78]]}, {\"category\": \"car\", \"corners_3d\": [[497.86, 233.79], [444.81, 234.11], [463.95, 223.78], [508.01, 223.57], [497.86, 191.03], [444.81, 191.13], [463.95, 188.05], [508.01, 187.98]]}]\n```" - } - ] - }, - { - "id": 64, - "data_source": "Mono3DRefer", - "scene_name": "000283", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[158.02, 264.69], [169.41, 257.48], [363.43, 249.81], [368.07, 255.72], [158.02, 192.72], [169.41, 191.16], [363.43, 189.5], [368.07, 190.78]]}, {\"category\": \"car\", \"corners_3d\": [[497.86, 233.79], [444.81, 234.11], [463.95, 223.78], [508.01, 223.57], [497.86, 191.03], [444.81, 191.13], [463.95, 188.05], [508.01, 187.98]]}, {\"category\": \"van\", \"corners_3d\": [[549.12, 192.02], [573.2, 191.67], [607.18, 193.42], [581.49, 193.83], [549.12, 156.06], [573.2, 156.36], [607.18, 154.83], [581.49, 154.47]]}]\n```" - } - ] - }, - { - "id": 65, - "data_source": "Mono3DRefer", - "scene_name": "000305", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[559.2, 191.01], [579.78, 191.02], [576.21, 191.7], [554.17, 191.69], [559.2, 171.44], [579.78, 171.43], [576.21, 170.71], [554.17, 170.73]]}, {\"category\": \"car\", \"corners_3d\": [[561.06, 197.23], [589.31, 197.28], [581.27, 199.11], [549.78, 199.05], [561.06, 168.34], [589.31, 168.3], [581.27, 166.76], [549.78, 166.81]]}, {\"category\": \"car\", \"corners_3d\": [[417.14, 222.97], [437.55, 225.23], [314.8, 230.16], [299.24, 227.38], [417.14, 179.44], [437.55, 179.32], [314.8, 179.08], [299.24, 179.22]]}, {\"category\": \"car\", \"corners_3d\": [[126.47, 236.19], [112.08, 240.81], [-116.58, 247.01], [-81.28, 241.42], [126.47, 183.25], [112.08, 183.39], [-116.58, 183.59], [-81.28, 183.41]]}, {\"category\": \"car\", \"corners_3d\": [[359.92, 204.31], [329.17, 204.14], [369.41, 201.79], [397.2, 201.92], [359.92, 170.96], [329.17, 171.04], [369.41, 172.13], [397.2, 172.07]]}]\n```" - } - ] - }, - { - "id": 66, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[551.01, 223.76], [510.28, 223.53], [534.86, 215.57], [569.27, 215.74], [551.01, 161.32], [510.28, 161.37], [534.86, 163.17], [569.27, 163.14]]}]\n```" - } - ] - }, - { - "id": 67, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[600.76, 219.67], [609.46, 219.23], [615.25, 219.7], [606.51, 220.15], [600.76, 182.39], [609.46, 182.3], [615.25, 182.39], [606.51, 182.49]]}]\n```" - } - ] - }, - { - "id": 68, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[363.53, 339.16], [234.74, 339.77], [339.73, 291.07], [430.76, 290.76], [363.53, 234.06], [234.74, 234.29], [339.73, 216.36], [430.76, 216.25]]}, {\"category\": \"car\", \"corners_3d\": [[460.99, 279.45], [379.65, 279.57], [426.4, 257.12], [490.59, 257.04], [460.99, 199.37], [379.65, 199.4], [426.4, 193.81], [490.59, 193.79]]}, {\"category\": \"car\", \"corners_3d\": [[611.24, 220.12], [643.76, 220.06], [649.45, 224.92], [613.59, 224.99], [611.24, 189.03], [643.76, 189.01], [649.45, 190.67], [613.59, 190.7]]}]\n```" - } - ] - }, - { - "id": 69, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[551.01, 223.76], [510.28, 223.53], [534.86, 215.57], [569.27, 215.74], [551.01, 161.32], [510.28, 161.37], [534.86, 163.17], [569.27, 163.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[600.76, 219.67], [609.46, 219.23], [615.25, 219.7], [606.51, 220.15], [600.76, 182.39], [609.46, 182.3], [615.25, 182.39], [606.51, 182.49]]}]\n```" - } - ] - }, - { - "id": 70, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[363.53, 339.16], [234.74, 339.77], [339.73, 291.07], [430.76, 290.76], [363.53, 234.06], [234.74, 234.29], [339.73, 216.36], [430.76, 216.25]]}, {\"category\": \"car\", \"corners_3d\": [[460.99, 279.45], [379.65, 279.57], [426.4, 257.12], [490.59, 257.04], [460.99, 199.37], [379.65, 199.4], [426.4, 193.81], [490.59, 193.79]]}, {\"category\": \"truck\", \"corners_3d\": [[551.01, 223.76], [510.28, 223.53], [534.86, 215.57], [569.27, 215.74], [551.01, 161.32], [510.28, 161.37], [534.86, 163.17], [569.27, 163.14]]}, {\"category\": \"car\", \"corners_3d\": [[611.24, 220.12], [643.76, 220.06], [649.45, 224.92], [613.59, 224.99], [611.24, 189.03], [643.76, 189.01], [649.45, 190.67], [613.59, 190.7]]}]\n```" - } - ] - }, - { - "id": 71, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[363.53, 339.16], [234.74, 339.77], [339.73, 291.07], [430.76, 290.76], [363.53, 234.06], [234.74, 234.29], [339.73, 216.36], [430.76, 216.25]]}, {\"category\": \"car\", \"corners_3d\": [[460.99, 279.45], [379.65, 279.57], [426.4, 257.12], [490.59, 257.04], [460.99, 199.37], [379.65, 199.4], [426.4, 193.81], [490.59, 193.79]]}, {\"category\": \"car\", \"corners_3d\": [[611.24, 220.12], [643.76, 220.06], [649.45, 224.92], [613.59, 224.99], [611.24, 189.03], [643.76, 189.01], [649.45, 190.67], [613.59, 190.7]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[600.76, 219.67], [609.46, 219.23], [615.25, 219.7], [606.51, 220.15], [600.76, 182.39], [609.46, 182.3], [615.25, 182.39], [606.51, 182.49]]}]\n```" - } - ] - }, - { - "id": 72, - "data_source": "Mono3DRefer", - "scene_name": "000312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[363.53, 339.16], [234.74, 339.77], [339.73, 291.07], [430.76, 290.76], [363.53, 234.06], [234.74, 234.29], [339.73, 216.36], [430.76, 216.25]]}, {\"category\": \"car\", \"corners_3d\": [[460.99, 279.45], [379.65, 279.57], [426.4, 257.12], [490.59, 257.04], [460.99, 199.37], [379.65, 199.4], [426.4, 193.81], [490.59, 193.79]]}, {\"category\": \"truck\", \"corners_3d\": [[551.01, 223.76], [510.28, 223.53], [534.86, 215.57], [569.27, 215.74], [551.01, 161.32], [510.28, 161.37], [534.86, 163.17], [569.27, 163.14]]}, {\"category\": \"car\", \"corners_3d\": [[611.24, 220.12], [643.76, 220.06], [649.45, 224.92], [613.59, 224.99], [611.24, 189.03], [643.76, 189.01], [649.45, 190.67], [613.59, 190.7]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[600.76, 219.67], [609.46, 219.23], [615.25, 219.7], [606.51, 220.15], [600.76, 182.39], [609.46, 182.3], [615.25, 182.39], [606.51, 182.49]]}]\n```" - } - ] - }, - { - "id": 73, - "data_source": "Mono3DRefer", - "scene_name": "000328", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-1148.49, 1382.39], [-2211.15, 1383.8], [29.35, 421.75], [247.42, 421.69], [-1148.49, 466.89], [-2211.15, 467.23], [29.35, 233.36], [247.42, 233.34]]}, {\"category\": \"car\", \"corners_3d\": [[293.82, 369.64], [129.62, 371.44], [282.14, 304.21], [390.16, 303.42], [293.82, 215.97], [129.62, 216.36], [282.14, 201.63], [390.16, 201.46]]}, {\"category\": \"car\", \"corners_3d\": [[424.82, 289.97], [323.88, 289.82], [396.91, 260.6], [472.68, 260.68], [424.82, 201.96], [323.88, 201.92], [396.91, 194.66], [472.68, 194.68]]}, {\"category\": \"car\", \"corners_3d\": [[761.34, 283.08], [870.13, 282.76], [1075.17, 364.87], [885.74, 365.85], [761.34, 183.1], [870.13, 183.07], [1075.17, 190.71], [885.74, 190.8]]}, {\"category\": \"car\", \"corners_3d\": [[718.52, 247.77], [791.14, 247.77], [843.15, 269.3], [749.66, 269.29], [718.52, 178.89], [791.14, 178.89], [843.15, 180.62], [749.66, 180.62]]}, {\"category\": \"car\", \"corners_3d\": [[985.47, 222.18], [964.85, 218.24], [1105.44, 219.58], [1139.32, 223.76], [985.47, 149.93], [964.85, 151.76], [1105.44, 151.14], [1139.32, 149.2]]}, {\"category\": \"car\", \"corners_3d\": [[641.46, 207.95], [670.69, 207.72], [688.21, 211.18], [656.2, 211.45], [641.46, 178.34], [670.69, 178.3], [688.21, 178.84], [656.2, 178.89]]}, {\"category\": \"car\", \"corners_3d\": [[589.26, 201.36], [609.66, 201.08], [626.54, 202.95], [604.95, 203.27], [589.26, 181.84], [609.66, 181.75], [626.54, 182.34], [604.95, 182.44]]}, {\"category\": \"car\", \"corners_3d\": [[561.79, 197.26], [551.89, 196.87], [583.51, 195.96], [593.52, 196.31], [561.79, 179.47], [551.89, 179.36], [583.51, 179.11], [593.52, 179.21]]}]\n```" - } - ] - }, - { - "id": 74, - "data_source": "Mono3DRefer", - "scene_name": "000328", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[501.81, 201.41], [493.31, 201.35], [499.95, 200.56], [508.23, 200.61], [501.81, 178.53], [493.31, 178.52], [499.95, 178.36], [508.23, 178.38]]}]\n```" - } - ] - }, - { - "id": 75, - "data_source": "Mono3DRefer", - "scene_name": "000328", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-1148.49, 1382.39], [-2211.15, 1383.8], [29.35, 421.75], [247.42, 421.69], [-1148.49, 466.89], [-2211.15, 467.23], [29.35, 233.36], [247.42, 233.34]]}, {\"category\": \"car\", \"corners_3d\": [[293.82, 369.64], [129.62, 371.44], [282.14, 304.21], [390.16, 303.42], [293.82, 215.97], [129.62, 216.36], [282.14, 201.63], [390.16, 201.46]]}, {\"category\": \"car\", \"corners_3d\": [[424.82, 289.97], [323.88, 289.82], [396.91, 260.6], [472.68, 260.68], [424.82, 201.96], [323.88, 201.92], [396.91, 194.66], [472.68, 194.68]]}, {\"category\": \"car\", \"corners_3d\": [[761.34, 283.08], [870.13, 282.76], [1075.17, 364.87], [885.74, 365.85], [761.34, 183.1], [870.13, 183.07], [1075.17, 190.71], [885.74, 190.8]]}, {\"category\": \"car\", \"corners_3d\": [[718.52, 247.77], [791.14, 247.77], [843.15, 269.3], [749.66, 269.29], [718.52, 178.89], [791.14, 178.89], [843.15, 180.62], [749.66, 180.62]]}, {\"category\": \"car\", \"corners_3d\": [[985.47, 222.18], [964.85, 218.24], [1105.44, 219.58], [1139.32, 223.76], [985.47, 149.93], [964.85, 151.76], [1105.44, 151.14], [1139.32, 149.2]]}, {\"category\": \"car\", \"corners_3d\": [[641.46, 207.95], [670.69, 207.72], [688.21, 211.18], [656.2, 211.45], [641.46, 178.34], [670.69, 178.3], [688.21, 178.84], [656.2, 178.89]]}, {\"category\": \"car\", \"corners_3d\": [[589.26, 201.36], [609.66, 201.08], [626.54, 202.95], [604.95, 203.27], [589.26, 181.84], [609.66, 181.75], [626.54, 182.34], [604.95, 182.44]]}, {\"category\": \"car\", \"corners_3d\": [[561.79, 197.26], [551.89, 196.87], [583.51, 195.96], [593.52, 196.31], [561.79, 179.47], [551.89, 179.36], [583.51, 179.11], [593.52, 179.21]]}, {\"category\": \"cyclist\", \"corners_3d\": [[501.81, 201.41], [493.31, 201.35], [499.95, 200.56], [508.23, 200.61], [501.81, 178.53], [493.31, 178.52], [499.95, 178.36], [508.23, 178.38]]}]\n```" - } - ] - }, - { - "id": 76, - "data_source": "Mono3DRefer", - "scene_name": "000346", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[725.66, 211.87], [755.51, 210.88], [817.42, 215.17], [785.91, 216.4], [725.66, 176.8], [755.51, 176.7], [817.42, 177.13], [785.91, 177.25]]}, {\"category\": \"car\", \"corners_3d\": [[625.78, 201.24], [651.31, 201.1], [663.92, 203.6], [636.19, 203.76], [625.78, 176.98], [651.31, 176.96], [663.92, 177.32], [636.19, 177.34]]}, {\"category\": \"car\", \"corners_3d\": [[929.02, 434.14], [1173.33, 430.61], [3987.18, 1655.62], [2721.11, 1780.62], [929.02, 181.92], [1173.33, 181.79], [3987.18, 224.28], [2721.11, 228.62]]}, {\"category\": \"car\", \"corners_3d\": [[-412.78, 363.37], [-204.55, 323.97], [163.53, 322.34], [52.8, 360.79], [-412.78, 169.26], [-204.55, 170.0], [163.53, 170.03], [52.8, 169.31]]}, {\"category\": \"car\", \"corners_3d\": [[350.69, 285.78], [296.36, 307.65], [-0.18, 305.97], [101.18, 284.6], [350.69, 180.45], [296.36, 181.92], [-0.18, 181.81], [101.18, 180.37]]}, {\"category\": \"car\", \"corners_3d\": [[99.02, 276.1], [163.23, 263.67], [341.02, 264.39], [300.78, 277.03], [99.02, 183.67], [163.23, 182.37], [341.02, 182.44], [300.78, 183.77]]}, {\"category\": \"car\", \"corners_3d\": [[276.62, 236.55], [302.85, 231.86], [418.76, 232.16], [401.66, 236.91], [276.62, 175.19], [302.85, 175.01], [418.76, 175.02], [401.66, 175.2]]}, {\"category\": \"car\", \"corners_3d\": [[311.51, 230.48], [332.72, 226.36], [446.37, 226.35], [433.9, 230.46], [311.51, 179.91], [332.72, 179.41], [446.37, 179.41], [433.9, 179.91]]}]\n```" - } - ] - }, - { - "id": 77, - "data_source": "Mono3DRefer", - "scene_name": "000346", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[340.45, 224.6], [359.1, 220.9], [485.23, 220.8], [476.32, 224.49], [340.45, 163.0], [359.1, 163.7], [485.23, 163.72], [476.32, 163.02]]}]\n```" - } - ] - }, - { - "id": 78, - "data_source": "Mono3DRefer", - "scene_name": "000346", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[725.66, 211.87], [755.51, 210.88], [817.42, 215.17], [785.91, 216.4], [725.66, 176.8], [755.51, 176.7], [817.42, 177.13], [785.91, 177.25]]}, {\"category\": \"car\", \"corners_3d\": [[625.78, 201.24], [651.31, 201.1], [663.92, 203.6], [636.19, 203.76], [625.78, 176.98], [651.31, 176.96], [663.92, 177.32], [636.19, 177.34]]}, {\"category\": \"car\", \"corners_3d\": [[929.02, 434.14], [1173.33, 430.61], [3987.18, 1655.62], [2721.11, 1780.62], [929.02, 181.92], [1173.33, 181.79], [3987.18, 224.28], [2721.11, 228.62]]}, {\"category\": \"car\", \"corners_3d\": [[-412.78, 363.37], [-204.55, 323.97], [163.53, 322.34], [52.8, 360.79], [-412.78, 169.26], [-204.55, 170.0], [163.53, 170.03], [52.8, 169.31]]}, {\"category\": \"car\", \"corners_3d\": [[350.69, 285.78], [296.36, 307.65], [-0.18, 305.97], [101.18, 284.6], [350.69, 180.45], [296.36, 181.92], [-0.18, 181.81], [101.18, 180.37]]}, {\"category\": \"car\", \"corners_3d\": [[99.02, 276.1], [163.23, 263.67], [341.02, 264.39], [300.78, 277.03], [99.02, 183.67], [163.23, 182.37], [341.02, 182.44], [300.78, 183.77]]}, {\"category\": \"car\", \"corners_3d\": [[276.62, 236.55], [302.85, 231.86], [418.76, 232.16], [401.66, 236.91], [276.62, 175.19], [302.85, 175.01], [418.76, 175.02], [401.66, 175.2]]}, {\"category\": \"car\", \"corners_3d\": [[311.51, 230.48], [332.72, 226.36], [446.37, 226.35], [433.9, 230.46], [311.51, 179.91], [332.72, 179.41], [446.37, 179.41], [433.9, 179.91]]}, {\"category\": \"van\", \"corners_3d\": [[340.45, 224.6], [359.1, 220.9], [485.23, 220.8], [476.32, 224.49], [340.45, 163.0], [359.1, 163.7], [485.23, 163.72], [476.32, 163.02]]}]\n```" - } - ] - }, - { - "id": 79, - "data_source": "Mono3DRefer", - "scene_name": "000357", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[509.79, 202.19], [543.47, 202.18], [536.42, 204.28], [498.58, 204.29], [509.79, 161.74], [543.47, 161.75], [536.42, 158.86], [498.58, 158.85]]}, {\"category\": \"van\", \"corners_3d\": [[543.58, 190.79], [561.23, 190.79], [559.8, 191.24], [540.71, 191.24], [543.58, 168.8], [561.23, 168.82], [559.8, 167.49], [540.71, 167.47]]}]\n```" - } - ] - }, - { - "id": 80, - "data_source": "Mono3DRefer", - "scene_name": "000357", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[575.48, 206.12], [609.57, 206.1], [611.44, 208.5], [573.42, 208.53], [575.48, 175.12], [609.57, 175.13], [611.44, 173.97], [573.42, 173.95]]}, {\"category\": \"car\", \"corners_3d\": [[-0.39, 288.15], [107.3, 288.01], [-146.53, 340.91], [-310.27, 341.23], [-0.39, 181.15], [107.3, 181.16], [-146.53, 179.07], [-310.27, 179.06]]}, {\"category\": \"car\", \"corners_3d\": [[297.91, 272.81], [391.56, 272.93], [287.19, 313.91], [150.1, 313.64], [297.91, 190.88], [391.56, 190.89], [287.19, 193.54], [150.1, 193.52]]}, {\"category\": \"car\", \"corners_3d\": [[271.97, 237.09], [331.79, 237.09], [261.01, 250.4], [185.84, 250.39], [271.97, 186.62], [331.79, 186.62], [261.01, 186.98], [185.84, 186.98]]}, {\"category\": \"car\", \"corners_3d\": [[431.94, 230.45], [483.81, 230.45], [458.24, 239.78], [395.67, 239.78], [431.94, 185.84], [483.81, 185.84], [458.24, 185.97], [395.67, 185.97]]}, {\"category\": \"car\", \"corners_3d\": [[367.62, 216.75], [406.08, 216.74], [378.66, 221.18], [334.76, 221.2], [367.62, 181.79], [406.08, 181.79], [378.66, 181.31], [334.76, 181.31]]}, {\"category\": \"car\", \"corners_3d\": [[457.96, 215.38], [497.31, 215.36], [481.66, 219.93], [436.34, 219.95], [457.96, 179.68], [497.31, 179.68], [481.66, 178.84], [436.34, 178.84]]}, {\"category\": \"car\", \"corners_3d\": [[158.08, 235.59], [216.08, 235.55], [129.03, 246.95], [57.8, 247.0], [158.08, 179.96], [216.08, 179.97], [129.03, 178.78], [57.8, 178.77]]}]\n```" - } - ] - }, - { - "id": 81, - "data_source": "Mono3DRefer", - "scene_name": "000357", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[575.48, 206.12], [609.57, 206.1], [611.44, 208.5], [573.42, 208.53], [575.48, 175.12], [609.57, 175.13], [611.44, 173.97], [573.42, 173.95]]}, {\"category\": \"car\", \"corners_3d\": [[-0.39, 288.15], [107.3, 288.01], [-146.53, 340.91], [-310.27, 341.23], [-0.39, 181.15], [107.3, 181.16], [-146.53, 179.07], [-310.27, 179.06]]}, {\"category\": \"car\", \"corners_3d\": [[297.91, 272.81], [391.56, 272.93], [287.19, 313.91], [150.1, 313.64], [297.91, 190.88], [391.56, 190.89], [287.19, 193.54], [150.1, 193.52]]}, {\"category\": \"car\", \"corners_3d\": [[271.97, 237.09], [331.79, 237.09], [261.01, 250.4], [185.84, 250.39], [271.97, 186.62], [331.79, 186.62], [261.01, 186.98], [185.84, 186.98]]}, {\"category\": \"car\", \"corners_3d\": [[431.94, 230.45], [483.81, 230.45], [458.24, 239.78], [395.67, 239.78], [431.94, 185.84], [483.81, 185.84], [458.24, 185.97], [395.67, 185.97]]}, {\"category\": \"car\", \"corners_3d\": [[367.62, 216.75], [406.08, 216.74], [378.66, 221.18], [334.76, 221.2], [367.62, 181.79], [406.08, 181.79], [378.66, 181.31], [334.76, 181.31]]}, {\"category\": \"car\", \"corners_3d\": [[457.96, 215.38], [497.31, 215.36], [481.66, 219.93], [436.34, 219.95], [457.96, 179.68], [497.31, 179.68], [481.66, 178.84], [436.34, 178.84]]}, {\"category\": \"van\", \"corners_3d\": [[509.79, 202.19], [543.47, 202.18], [536.42, 204.28], [498.58, 204.29], [509.79, 161.74], [543.47, 161.75], [536.42, 158.86], [498.58, 158.85]]}, {\"category\": \"van\", \"corners_3d\": [[543.58, 190.79], [561.23, 190.79], [559.8, 191.24], [540.71, 191.24], [543.58, 168.8], [561.23, 168.82], [559.8, 167.49], [540.71, 167.47]]}, {\"category\": \"car\", \"corners_3d\": [[158.08, 235.59], [216.08, 235.55], [129.03, 246.95], [57.8, 247.0], [158.08, 179.96], [216.08, 179.97], [129.03, 178.78], [57.8, 178.77]]}]\n```" - } - ] - }, - { - "id": 82, - "data_source": "Mono3DRefer", - "scene_name": "000362", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[697.4, 255.03], [780.22, 254.66], [843.55, 280.7], [734.59, 281.35], [697.4, 182.13], [780.22, 182.09], [843.55, 185.03], [734.59, 185.1]]}, {\"category\": \"car\", \"corners_3d\": [[418.78, 241.39], [482.38, 240.98], [467.08, 254.49], [390.67, 255.08], [418.78, 193.02], [482.38, 192.9], [467.08, 196.88], [390.67, 197.05]]}, {\"category\": \"car\", \"corners_3d\": [[653.75, 225.2], [698.59, 225.07], [716.96, 233.25], [665.14, 233.42], [653.75, 182.44], [698.59, 182.42], [716.96, 183.92], [665.14, 183.95]]}, {\"category\": \"car\", \"corners_3d\": [[472.6, 222.38], [515.26, 222.38], [500.24, 230.22], [450.84, 230.22], [472.6, 181.61], [515.26, 181.62], [500.24, 183.0], [450.84, 183.0]]}, {\"category\": \"car\", \"corners_3d\": [[526.61, 213.07], [496.84, 213.1], [505.28, 209.63], [532.46, 209.6], [526.61, 184.11], [496.84, 184.12], [505.28, 183.15], [532.46, 183.14]]}, {\"category\": \"car\", \"corners_3d\": [[631.74, 207.56], [661.25, 207.52], [668.33, 210.9], [635.96, 210.95], [631.74, 179.05], [661.25, 179.04], [668.33, 179.64], [635.96, 179.65]]}]\n```" - } - ] - }, - { - "id": 83, - "data_source": "Mono3DRefer", - "scene_name": "000362", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[574.19, 198.77], [567.89, 198.77], [569.24, 198.04], [575.36, 198.04], [574.19, 180.26], [567.89, 180.25], [569.24, 180.05], [575.36, 180.05]]}, {\"category\": \"cyclist\", \"corners_3d\": [[581.83, 197.39], [575.25, 197.39], [576.08, 196.79], [582.5, 196.79], [581.83, 179.83], [575.25, 179.83], [576.08, 179.66], [582.5, 179.66]]}]\n```" - } - ] - }, - { - "id": 84, - "data_source": "Mono3DRefer", - "scene_name": "000362", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[697.4, 255.03], [780.22, 254.66], [843.55, 280.7], [734.59, 281.35], [697.4, 182.13], [780.22, 182.09], [843.55, 185.03], [734.59, 185.1]]}, {\"category\": \"car\", \"corners_3d\": [[418.78, 241.39], [482.38, 240.98], [467.08, 254.49], [390.67, 255.08], [418.78, 193.02], [482.38, 192.9], [467.08, 196.88], [390.67, 197.05]]}, {\"category\": \"car\", \"corners_3d\": [[653.75, 225.2], [698.59, 225.07], [716.96, 233.25], [665.14, 233.42], [653.75, 182.44], [698.59, 182.42], [716.96, 183.92], [665.14, 183.95]]}, {\"category\": \"car\", \"corners_3d\": [[472.6, 222.38], [515.26, 222.38], [500.24, 230.22], [450.84, 230.22], [472.6, 181.61], [515.26, 181.62], [500.24, 183.0], [450.84, 183.0]]}, {\"category\": \"car\", \"corners_3d\": [[526.61, 213.07], [496.84, 213.1], [505.28, 209.63], [532.46, 209.6], [526.61, 184.11], [496.84, 184.12], [505.28, 183.15], [532.46, 183.14]]}, {\"category\": \"cyclist\", \"corners_3d\": [[574.19, 198.77], [567.89, 198.77], [569.24, 198.04], [575.36, 198.04], [574.19, 180.26], [567.89, 180.25], [569.24, 180.05], [575.36, 180.05]]}, {\"category\": \"cyclist\", \"corners_3d\": [[581.83, 197.39], [575.25, 197.39], [576.08, 196.79], [582.5, 196.79], [581.83, 179.83], [575.25, 179.83], [576.08, 179.66], [582.5, 179.66]]}, {\"category\": \"car\", \"corners_3d\": [[631.74, 207.56], [661.25, 207.52], [668.33, 210.9], [635.96, 210.95], [631.74, 179.05], [661.25, 179.04], [668.33, 179.64], [635.96, 179.65]]}]\n```" - } - ] - }, - { - "id": 85, - "data_source": "Mono3DRefer", - "scene_name": "000373", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[325.2, 232.71], [319.34, 237.27], [165.91, 240.8], [183.5, 235.74], [325.2, 173.84], [319.34, 173.92], [165.91, 173.97], [183.5, 173.89]]}, {\"category\": \"car\", \"corners_3d\": [[480.15, 215.45], [512.28, 214.81], [530.71, 219.48], [495.26, 220.28], [480.15, 182.86], [512.28, 182.71], [530.71, 183.81], [495.26, 183.99]]}, {\"category\": \"car\", \"corners_3d\": [[-127.77, 294.19], [-41.77, 277.36], [206.51, 272.25], [163.11, 287.36], [-127.77, 167.12], [-41.77, 167.91], [206.51, 168.15], [163.11, 167.44]]}, {\"category\": \"car\", \"corners_3d\": [[13.48, 266.88], [70.67, 255.81], [264.91, 252.75], [234.94, 262.96], [13.48, 183.96], [70.67, 182.66], [264.91, 182.29], [234.94, 183.5]]}, {\"category\": \"car\", \"corners_3d\": [[105.93, 250.26], [140.96, 243.23], [298.64, 240.97], [280.03, 247.52], [105.93, 183.96], [140.96, 182.95], [298.64, 182.63], [280.03, 183.57]]}]\n```" - } - ] - }, - { - "id": 86, - "data_source": "Mono3DRefer", - "scene_name": "000378", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[405.98, 233.87], [367.31, 231.17], [468.36, 223.04], [505.42, 225.02], [405.98, 182.61], [367.31, 182.18], [468.36, 180.88], [505.42, 181.19]]}, {\"category\": \"car\", \"corners_3d\": [[608.27, 215.84], [581.62, 213.42], [678.19, 209.94], [707.67, 211.95], [608.27, 175.05], [581.62, 174.93], [678.19, 174.75], [707.67, 174.85]]}, {\"category\": \"car\", \"corners_3d\": [[772.29, 229.04], [788.62, 234.21], [622.29, 234.5], [619.99, 229.28], [772.29, 170.32], [788.62, 170.09], [622.29, 170.07], [619.99, 170.31]]}]\n```" - } - ] - }, - { - "id": 87, - "data_source": "Mono3DRefer", - "scene_name": "000401", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[127.97, 394.72], [354.96, 396.19], [-593.06, 1181.54], [-1567.45, 1152.19], [127.97, 195.95], [354.96, 196.1], [-593.06, 277.86], [-1567.45, 274.8]]}, {\"category\": \"car\", \"corners_3d\": [[659.82, 234.25], [721.55, 233.89], [766.88, 250.96], [688.09, 251.56], [659.82, 177.42], [721.55, 177.39], [766.88, 178.66], [688.09, 178.7]]}, {\"category\": \"car\", \"corners_3d\": [[292.91, 290.58], [415.3, 289.43], [302.94, 375.89], [86.25, 379.41], [292.91, 185.26], [415.3, 185.14], [302.94, 194.25], [86.25, 194.62]]}, {\"category\": \"car\", \"corners_3d\": [[517.56, 209.79], [556.84, 209.67], [555.49, 214.7], [510.81, 214.85], [517.56, 173.97], [556.84, 173.96], [555.49, 174.11], [510.81, 174.12]]}, {\"category\": \"car\", \"corners_3d\": [[635.1, 210.34], [672.61, 210.16], [691.36, 215.65], [648.41, 215.88], [635.1, 173.54], [672.61, 173.54], [691.36, 173.64], [648.41, 173.64]]}, {\"category\": \"car\", \"corners_3d\": [[542.63, 201.58], [567.19, 201.61], [562.31, 203.78], [535.9, 203.74], [542.63, 178.53], [567.19, 178.54], [562.31, 178.97], [535.9, 178.96]]}, {\"category\": \"car\", \"corners_3d\": [[622.82, 198.26], [648.85, 198.28], [651.34, 200.89], [622.64, 200.87], [622.82, 171.07], [648.85, 171.06], [651.34, 170.88], [622.64, 170.88]]}]\n```" - } - ] - }, - { - "id": 88, - "data_source": "Mono3DRefer", - "scene_name": "000401", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[128.73, 334.39], [186.75, 334.4], [163.38, 343.31], [102.16, 343.3], [128.73, 157.34], [186.75, 157.34], [163.38, 156.48], [102.16, 156.48]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.97, 297.62], [252.49, 297.62], [240.67, 301.75], [215.34, 301.74], [227.97, 167.06], [252.49, 167.06], [240.67, 166.87], [215.34, 166.87]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[255.75, 303.84], [285.34, 303.85], [270.35, 309.89], [239.39, 309.89], [255.75, 172.09], [285.34, 172.09], [270.35, 172.05], [239.39, 172.05]]}]\n```" - } - ] - }, - { - "id": 89, - "data_source": "Mono3DRefer", - "scene_name": "000401", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[127.97, 394.72], [354.96, 396.19], [-593.06, 1181.54], [-1567.45, 1152.19], [127.97, 195.95], [354.96, 196.1], [-593.06, 277.86], [-1567.45, 274.8]]}, {\"category\": \"car\", \"corners_3d\": [[659.82, 234.25], [721.55, 233.89], [766.88, 250.96], [688.09, 251.56], [659.82, 177.42], [721.55, 177.39], [766.88, 178.66], [688.09, 178.7]]}, {\"category\": \"car\", \"corners_3d\": [[292.91, 290.58], [415.3, 289.43], [302.94, 375.89], [86.25, 379.41], [292.91, 185.26], [415.3, 185.14], [302.94, 194.25], [86.25, 194.62]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[128.73, 334.39], [186.75, 334.4], [163.38, 343.31], [102.16, 343.3], [128.73, 157.34], [186.75, 157.34], [163.38, 156.48], [102.16, 156.48]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.97, 297.62], [252.49, 297.62], [240.67, 301.75], [215.34, 301.74], [227.97, 167.06], [252.49, 167.06], [240.67, 166.87], [215.34, 166.87]]}, {\"category\": \"car\", \"corners_3d\": [[517.56, 209.79], [556.84, 209.67], [555.49, 214.7], [510.81, 214.85], [517.56, 173.97], [556.84, 173.96], [555.49, 174.11], [510.81, 174.12]]}, {\"category\": \"car\", \"corners_3d\": [[635.1, 210.34], [672.61, 210.16], [691.36, 215.65], [648.41, 215.88], [635.1, 173.54], [672.61, 173.54], [691.36, 173.64], [648.41, 173.64]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[255.75, 303.84], [285.34, 303.85], [270.35, 309.89], [239.39, 309.89], [255.75, 172.09], [285.34, 172.09], [270.35, 172.05], [239.39, 172.05]]}, {\"category\": \"car\", \"corners_3d\": [[542.63, 201.58], [567.19, 201.61], [562.31, 203.78], [535.9, 203.74], [542.63, 178.53], [567.19, 178.54], [562.31, 178.97], [535.9, 178.96]]}, {\"category\": \"car\", \"corners_3d\": [[622.82, 198.26], [648.85, 198.28], [651.34, 200.89], [622.64, 200.87], [622.82, 171.07], [648.85, 171.06], [651.34, 170.88], [622.64, 170.88]]}]\n```" - } - ] - }, - { - "id": 90, - "data_source": "Mono3DRefer", - "scene_name": "000409", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[272.22, 208.45], [283.77, 208.42], [268.96, 210.2], [256.82, 210.22], [272.22, 174.74], [283.77, 174.74], [268.96, 174.83], [256.82, 174.83]]}]\n```" - } - ] - }, - { - "id": 91, - "data_source": "Mono3DRefer", - "scene_name": "000409", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[569.28, 283.54], [684.67, 283.56], [713.94, 327.03], [553.24, 327.0], [569.28, 172.83], [684.67, 172.83], [713.94, 172.82], [553.24, 172.82]]}, {\"category\": \"car\", \"corners_3d\": [[473.89, 194.74], [452.6, 194.76], [461.98, 193.23], [481.78, 193.22], [473.89, 176.66], [452.6, 176.67], [461.98, 176.4], [481.78, 176.4]]}]\n```" - } - ] - }, - { - "id": 92, - "data_source": "Mono3DRefer", - "scene_name": "000409", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[569.28, 283.54], [684.67, 283.56], [713.94, 327.03], [553.24, 327.0], [569.28, 172.83], [684.67, 172.83], [713.94, 172.82], [553.24, 172.82]]}, {\"category\": \"cyclist\", \"corners_3d\": [[272.22, 208.45], [283.77, 208.42], [268.96, 210.2], [256.82, 210.22], [272.22, 174.74], [283.77, 174.74], [268.96, 174.83], [256.82, 174.83]]}, {\"category\": \"car\", \"corners_3d\": [[473.89, 194.74], [452.6, 194.76], [461.98, 193.23], [481.78, 193.22], [473.89, 176.66], [452.6, 176.67], [461.98, 176.4], [481.78, 176.4]]}]\n```" - } - ] - }, - { - "id": 93, - "data_source": "Mono3DRefer", - "scene_name": "000413", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[583.02, 210.02], [615.16, 209.98], [618.03, 213.96], [582.43, 214.02], [583.02, 178.24], [615.16, 178.23], [618.03, 178.81], [582.43, 178.82]]}, {\"category\": \"car\", \"corners_3d\": [[193.58, 279.62], [102.51, 280.03], [204.91, 257.39], [276.49, 257.14], [193.58, 189.31], [102.51, 189.37], [204.91, 185.88], [276.49, 185.84]]}, {\"category\": \"car\", \"corners_3d\": [[408.82, 223.21], [364.43, 223.21], [397.2, 216.46], [435.64, 216.46], [408.82, 186.46], [364.43, 186.46], [397.2, 184.63], [435.64, 184.63]]}]\n```" - } - ] - }, - { - "id": 94, - "data_source": "Mono3DRefer", - "scene_name": "000428", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[451.55, 244.68], [427.46, 244.22], [443.13, 241.09], [466.14, 241.51], [451.55, 163.92], [427.46, 164.04], [443.13, 164.85], [466.14, 164.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[434.88, 246.84], [406.83, 246.55], [421.41, 243.04], [448.03, 243.31], [434.88, 166.32], [406.83, 166.39], [421.41, 167.14], [448.03, 167.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[411.47, 213.3], [393.64, 213.34], [398.84, 212.36], [416.13, 212.33], [411.47, 170.38], [393.64, 170.36], [398.84, 170.67], [416.13, 170.68]]}]\n```" - } - ] - }, - { - "id": 95, - "data_source": "Mono3DRefer", - "scene_name": "000428", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[544.34, 202.77], [577.07, 202.63], [586.1, 205.71], [548.84, 205.88], [544.34, 164.75], [577.07, 164.84], [586.1, 162.66], [548.84, 162.54]]}]\n```" - } - ] - }, - { - "id": 96, - "data_source": "Mono3DRefer", - "scene_name": "000428", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[544.34, 202.77], [577.07, 202.63], [586.1, 205.71], [548.84, 205.88], [544.34, 164.75], [577.07, 164.84], [586.1, 162.66], [548.84, 162.54]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[451.55, 244.68], [427.46, 244.22], [443.13, 241.09], [466.14, 241.51], [451.55, 163.92], [427.46, 164.04], [443.13, 164.85], [466.14, 164.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[434.88, 246.84], [406.83, 246.55], [421.41, 243.04], [448.03, 243.31], [434.88, 166.32], [406.83, 166.39], [421.41, 167.14], [448.03, 167.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[411.47, 213.3], [393.64, 213.34], [398.84, 212.36], [416.13, 212.33], [411.47, 170.38], [393.64, 170.36], [398.84, 170.67], [416.13, 170.68]]}]\n```" - } - ] - }, - { - "id": 97, - "data_source": "Mono3DRefer", - "scene_name": "000429", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[274.03, 330.57], [128.48, 330.59], [279.14, 281.13], [379.04, 281.12], [274.03, 201.74], [128.48, 201.75], [279.14, 192.69], [379.04, 192.69]]}, {\"category\": \"car\", \"corners_3d\": [[917.84, 344.43], [1083.42, 344.86], [1605.28, 537.45], [1252.55, 535.56], [917.84, 193.37], [1083.42, 193.42], [1605.28, 216.45], [1252.55, 216.22]]}, {\"category\": \"car\", \"corners_3d\": [[413.98, 264.69], [326.62, 264.59], [388.48, 244.99], [457.21, 245.05], [413.98, 190.06], [326.62, 190.04], [388.48, 186.37], [457.21, 186.38]]}, {\"category\": \"car\", \"corners_3d\": [[782.94, 267.91], [877.82, 268.05], [979.87, 305.34], [847.69, 305.08], [782.94, 182.05], [877.82, 182.07], [979.87, 185.68], [847.69, 185.65]]}, {\"category\": \"car\", \"corners_3d\": [[464.68, 239.98], [401.57, 239.69], [446.82, 227.27], [498.31, 227.47], [464.68, 187.34], [401.57, 187.27], [446.82, 184.59], [498.31, 184.64]]}, {\"category\": \"car\", \"corners_3d\": [[724.85, 237.26], [790.48, 237.26], [844.23, 256.46], [759.04, 256.45], [724.85, 177.24], [790.48, 177.24], [844.23, 178.55], [759.04, 178.55]]}, {\"category\": \"car\", \"corners_3d\": [[543.85, 208.99], [511.32, 208.95], [521.22, 205.95], [551.06, 205.99], [543.85, 179.38], [511.32, 179.37], [521.22, 178.83], [551.06, 178.84]]}, {\"category\": \"car\", \"corners_3d\": [[706.98, 224.5], [759.9, 224.62], [783.23, 234.04], [720.64, 233.88], [706.98, 176.65], [759.9, 176.65], [783.23, 177.35], [720.64, 177.33]]}, {\"category\": \"car\", \"corners_3d\": [[697.8, 214.19], [739.16, 214.29], [753.38, 220.17], [706.12, 220.05], [697.8, 175.65], [739.16, 175.66], [753.38, 176.06], [706.12, 176.05]]}, {\"category\": \"car\", \"corners_3d\": [[685.73, 208.38], [721.58, 208.42], [733.84, 212.91], [693.44, 212.86], [685.73, 176.68], [721.58, 176.68], [733.84, 177.16], [693.44, 177.16]]}]\n```" - } - ] - }, - { - "id": 98, - "data_source": "Mono3DRefer", - "scene_name": "000453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1034.22, 202.95], [1063.96, 204.67], [922.29, 204.39], [902.86, 202.71], [1034.22, 159.4], [1063.96, 157.78], [922.29, 158.05], [902.86, 159.63]]}, {\"category\": \"car\", \"corners_3d\": [[471.68, 199.06], [440.41, 199.27], [438.25, 197.65], [466.87, 197.48], [471.68, 172.3], [440.41, 172.2], [438.25, 172.92], [466.87, 173.0]]}, {\"category\": \"car\", \"corners_3d\": [[291.26, 214.02], [247.63, 214.18], [287.1, 209.74], [324.84, 209.61], [291.26, 175.5], [247.63, 175.48], [287.1, 176.14], [324.84, 176.16]]}, {\"category\": \"car\", \"corners_3d\": [[65.32, 408.28], [305.79, 409.13], [-446.5, 971.32], [-1261.32, 961.32], [65.32, 204.7], [305.79, 204.79], [-446.5, 264.5], [-1261.32, 263.43]]}]\n```" - } - ] - }, - { - "id": 99, - "data_source": "Mono3DRefer", - "scene_name": "000453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[992.53, 213.43], [1025.78, 214.06], [1029.58, 216.97], [993.43, 216.22], [992.53, 150.39], [1025.78, 149.81], [1029.58, 147.15], [993.43, 147.84]]}]\n```" - } - ] - }, - { - "id": 100, - "data_source": "Mono3DRefer", - "scene_name": "000453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[992.53, 213.43], [1025.78, 214.06], [1029.58, 216.97], [993.43, 216.22], [992.53, 150.39], [1025.78, 149.81], [1029.58, 147.15], [993.43, 147.84]]}, {\"category\": \"car\", \"corners_3d\": [[1034.22, 202.95], [1063.96, 204.67], [922.29, 204.39], [902.86, 202.71], [1034.22, 159.4], [1063.96, 157.78], [922.29, 158.05], [902.86, 159.63]]}, {\"category\": \"car\", \"corners_3d\": [[471.68, 199.06], [440.41, 199.27], [438.25, 197.65], [466.87, 197.48], [471.68, 172.3], [440.41, 172.2], [438.25, 172.92], [466.87, 173.0]]}, {\"category\": \"car\", \"corners_3d\": [[291.26, 214.02], [247.63, 214.18], [287.1, 209.74], [324.84, 209.61], [291.26, 175.5], [247.63, 175.48], [287.1, 176.14], [324.84, 176.16]]}, {\"category\": \"car\", \"corners_3d\": [[65.32, 408.28], [305.79, 409.13], [-446.5, 971.32], [-1261.32, 961.32], [65.32, 204.7], [305.79, 204.79], [-446.5, 264.5], [-1261.32, 263.43]]}]\n```" - } - ] - }, - { - "id": 101, - "data_source": "Mono3DRefer", - "scene_name": "000457", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[583.92, 203.63], [614.89, 203.61], [616.91, 206.85], [582.68, 206.88], [583.92, 179.24], [614.89, 179.23], [616.91, 179.9], [582.68, 179.91]]}, {\"category\": \"car\", \"corners_3d\": [[544.09, 201.76], [570.57, 201.75], [567.75, 204.27], [538.95, 204.28], [544.09, 179.51], [570.57, 179.5], [567.75, 180.09], [538.95, 180.09]]}]\n```" - } - ] - }, - { - "id": 102, - "data_source": "Mono3DRefer", - "scene_name": "000457", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[607.36, 190.52], [624.71, 190.52], [625.92, 191.98], [607.13, 191.98], [607.36, 166.61], [624.71, 166.61], [625.92, 166.09], [607.13, 166.09]]}]\n```" - } - ] - }, - { - "id": 103, - "data_source": "Mono3DRefer", - "scene_name": "000457", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[583.92, 203.63], [614.89, 203.61], [616.91, 206.85], [582.68, 206.88], [583.92, 179.24], [614.89, 179.23], [616.91, 179.9], [582.68, 179.91]]}, {\"category\": \"car\", \"corners_3d\": [[544.09, 201.76], [570.57, 201.75], [567.75, 204.27], [538.95, 204.28], [544.09, 179.51], [570.57, 179.5], [567.75, 180.09], [538.95, 180.09]]}, {\"category\": \"van\", \"corners_3d\": [[607.36, 190.52], [624.71, 190.52], [625.92, 191.98], [607.13, 191.98], [607.36, 166.61], [624.71, 166.61], [625.92, 166.09], [607.13, 166.09]]}]\n```" - } - ] - }, - { - "id": 104, - "data_source": "Mono3DRefer", - "scene_name": "000459", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[26.24, 559.18], [-345.89, 555.29], [175.04, 349.76], [348.68, 350.59], [26.24, 222.2], [-345.89, 221.7], [175.04, 195.45], [348.68, 195.56]]}, {\"category\": \"car\", \"corners_3d\": [[493.58, 211.84], [523.98, 211.73], [520.49, 215.54], [487.1, 215.68], [493.58, 181.23], [523.98, 181.21], [520.49, 182.03], [487.1, 182.05]]}, {\"category\": \"car\", \"corners_3d\": [[526.54, 208.61], [496.22, 208.65], [504.31, 205.47], [531.93, 205.43], [526.54, 180.57], [496.22, 180.58], [504.31, 179.9], [531.93, 179.89]]}, {\"category\": \"car\", \"corners_3d\": [[311.11, 220.27], [319.78, 217.55], [433.6, 216.22], [431.92, 218.77], [311.11, 174.4], [319.78, 174.31], [433.6, 174.27], [431.92, 174.35]]}]\n```" - } - ] - }, - { - "id": 105, - "data_source": "Mono3DRefer", - "scene_name": "000478", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[552.61, 192.58], [574.51, 192.57], [573.59, 193.41], [550.02, 193.42], [552.61, 171.81], [574.51, 171.81], [573.59, 171.08], [550.02, 171.07]]}, {\"category\": \"car\", \"corners_3d\": [[547.41, 203.23], [579.35, 203.42], [557.81, 206.37], [521.81, 206.13], [547.41, 169.91], [579.35, 169.81], [557.81, 168.23], [521.81, 168.36]]}, {\"category\": \"car\", \"corners_3d\": [[289.08, 232.93], [292.83, 236.53], [132.35, 241.66], [139.69, 237.38], [289.08, 184.07], [292.83, 184.25], [132.35, 184.5], [139.69, 184.29]]}, {\"category\": \"car\", \"corners_3d\": [[359.92, 204.11], [329.17, 203.94], [369.41, 201.61], [397.2, 201.74], [359.92, 170.76], [329.17, 170.84], [369.41, 171.95], [397.2, 171.89]]}]\n```" - } - ] - }, - { - "id": 106, - "data_source": "Mono3DRefer", - "scene_name": "000486", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[741.09, 213.4], [759.72, 213.4], [764.41, 214.67], [745.19, 214.67], [741.09, 159.69], [759.72, 159.69], [764.41, 159.28], [745.19, 159.28]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[761.14, 212.99], [779.13, 212.99], [784.07, 214.16], [765.55, 214.16], [761.14, 159.28], [779.13, 159.28], [784.07, 158.88], [765.55, 158.88]]}]\n```" - } - ] - }, - { - "id": 107, - "data_source": "Mono3DRefer", - "scene_name": "000486", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[688.2, 221.05], [731.02, 221.03], [752.03, 228.93], [702.2, 228.96], [688.2, 179.94], [731.02, 179.94], [752.03, 181.1], [702.2, 181.11]]}, {\"category\": \"car\", \"corners_3d\": [[935.58, 384.72], [1158.36, 382.11], [2735.88, 954.1], [1950.88, 991.8], [935.58, 183.81], [1158.36, 183.68], [2735.88, 213.27], [1950.88, 215.22]]}, {\"category\": \"car\", \"corners_3d\": [[390.2, 262.37], [317.8, 262.2], [373.76, 245.84], [432.96, 245.96], [390.2, 192.54], [317.8, 192.5], [373.76, 188.9], [432.96, 188.93]]}, {\"category\": \"car\", \"corners_3d\": [[728.86, 258.11], [813.39, 257.81], [892.6, 287.74], [778.53, 288.27], [728.86, 177.63], [813.39, 177.62], [892.6, 179.29], [778.53, 179.32]]}, {\"category\": \"car\", \"corners_3d\": [[695.9, 236.7], [751.86, 236.2], [797.2, 249.61], [729.71, 250.34], [695.9, 178.19], [751.86, 178.15], [797.2, 179.27], [729.71, 179.33]]}, {\"category\": \"car\", \"corners_3d\": [[445.32, 240.62], [396.71, 240.67], [425.82, 231.04], [467.52, 231.01], [445.32, 191.14], [396.71, 191.16], [425.82, 188.56], [467.52, 188.55]]}]\n```" - } - ] - }, - { - "id": 108, - "data_source": "Mono3DRefer", - "scene_name": "000486", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[741.09, 213.4], [759.72, 213.4], [764.41, 214.67], [745.19, 214.67], [741.09, 159.69], [759.72, 159.69], [764.41, 159.28], [745.19, 159.28]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[761.14, 212.99], [779.13, 212.99], [784.07, 214.16], [765.55, 214.16], [761.14, 159.28], [779.13, 159.28], [784.07, 158.88], [765.55, 158.88]]}, {\"category\": \"car\", \"corners_3d\": [[688.2, 221.05], [731.02, 221.03], [752.03, 228.93], [702.2, 228.96], [688.2, 179.94], [731.02, 179.94], [752.03, 181.1], [702.2, 181.11]]}, {\"category\": \"car\", \"corners_3d\": [[935.58, 384.72], [1158.36, 382.11], [2735.88, 954.1], [1950.88, 991.8], [935.58, 183.81], [1158.36, 183.68], [2735.88, 213.27], [1950.88, 215.22]]}, {\"category\": \"car\", \"corners_3d\": [[390.2, 262.37], [317.8, 262.2], [373.76, 245.84], [432.96, 245.96], [390.2, 192.54], [317.8, 192.5], [373.76, 188.9], [432.96, 188.93]]}, {\"category\": \"car\", \"corners_3d\": [[728.86, 258.11], [813.39, 257.81], [892.6, 287.74], [778.53, 288.27], [728.86, 177.63], [813.39, 177.62], [892.6, 179.29], [778.53, 179.32]]}, {\"category\": \"car\", \"corners_3d\": [[695.9, 236.7], [751.86, 236.2], [797.2, 249.61], [729.71, 250.34], [695.9, 178.19], [751.86, 178.15], [797.2, 179.27], [729.71, 179.33]]}, {\"category\": \"car\", \"corners_3d\": [[445.32, 240.62], [396.71, 240.67], [425.82, 231.04], [467.52, 231.01], [445.32, 191.14], [396.71, 191.16], [425.82, 188.56], [467.52, 188.55]]}]\n```" - } - ] - }, - { - "id": 109, - "data_source": "Mono3DRefer", - "scene_name": "000499", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[501.17, 201.98], [488.67, 201.97], [490.24, 201.64], [502.59, 201.64], [501.17, 177.55], [488.67, 177.55], [490.24, 177.5], [502.59, 177.5]]}]\n```" - } - ] - }, - { - "id": 110, - "data_source": "Mono3DRefer", - "scene_name": "000510", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-55.32, 334.86], [72.98, 311.06], [279.56, 325.65], [175.14, 355.26], [-55.32, 214.56], [72.98, 208.44], [279.56, 212.19], [175.14, 219.81]]}, {\"category\": \"car\", \"corners_3d\": [[670.89, 236.34], [737.6, 237.64], [729.04, 254.0], [646.09, 251.99], [670.89, 176.22], [737.6, 176.29], [729.04, 177.16], [646.09, 177.05]]}, {\"category\": \"car\", \"corners_3d\": [[885.65, 238.21], [960.04, 239.05], [1041.29, 260.1], [942.28, 258.64], [885.65, 169.83], [960.04, 169.79], [1041.29, 168.81], [942.28, 168.88]]}, {\"category\": \"car\", \"corners_3d\": [[861.65, 218.42], [921.27, 219.17], [958.7, 230.19], [884.43, 229.05], [861.65, 171.91], [921.27, 171.9], [958.7, 171.67], [884.43, 171.69]]}, {\"category\": \"car\", \"corners_3d\": [[721.45, 211.8], [757.01, 212.31], [751.32, 216.56], [712.06, 215.94], [721.45, 179.11], [757.01, 179.2], [751.32, 179.88], [712.06, 179.78]]}, {\"category\": \"car\", \"corners_3d\": [[841.25, 204.06], [881.43, 204.55], [888.67, 209.32], [842.41, 208.67], [841.25, 164.67], [881.43, 164.54], [888.67, 163.29], [842.41, 163.46]]}]\n```" - } - ] - }, - { - "id": 111, - "data_source": "Mono3DRefer", - "scene_name": "000558", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[94.45, 244.99], [161.02, 244.94], [42.96, 260.98], [-41.64, 261.06], [94.45, 181.18], [161.02, 181.19], [42.96, 180.1], [-41.64, 180.1]]}, {\"category\": \"car\", \"corners_3d\": [[575.2, 206.82], [609.8, 206.8], [611.73, 209.33], [573.07, 209.36], [575.2, 175.36], [609.8, 175.37], [611.73, 174.22], [573.07, 174.2]]}, {\"category\": \"car\", \"corners_3d\": [[218.01, 298.44], [336.63, 298.64], [148.1, 375.47], [-50.07, 374.9], [218.01, 194.58], [336.63, 194.59], [148.1, 200.94], [-50.07, 200.9]]}, {\"category\": \"car\", \"corners_3d\": [[222.31, 246.15], [291.14, 246.15], [193.94, 264.86], [103.99, 264.85], [222.31, 188.07], [291.14, 188.07], [193.94, 188.95], [103.99, 188.95]]}, {\"category\": \"car\", \"corners_3d\": [[408.41, 238.8], [467.69, 238.8], [433.66, 251.82], [359.98, 251.82], [408.41, 187.82], [467.69, 187.82], [433.66, 188.45], [359.98, 188.45]]}, {\"category\": \"car\", \"corners_3d\": [[345.96, 221.03], [388.34, 221.01], [354.9, 226.65], [305.82, 226.67], [345.96, 182.51], [388.34, 182.51], [354.9, 182.09], [305.82, 182.08]]}, {\"category\": \"car\", \"corners_3d\": [[444.42, 219.49], [487.76, 219.47], [468.65, 225.28], [417.94, 225.3], [444.42, 180.17], [487.76, 180.17], [468.65, 179.32], [417.94, 179.32]]}]\n```" - } - ] - }, - { - "id": 112, - "data_source": "Mono3DRefer", - "scene_name": "000558", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[504.42, 204.26], [540.39, 204.25], [532.39, 206.78], [491.63, 206.79], [504.42, 161.05], [540.39, 161.06], [532.39, 157.85], [491.63, 157.84]]}, {\"category\": \"van\", \"corners_3d\": [[542.58, 191.16], [560.84, 191.16], [559.32, 191.65], [539.52, 191.66], [542.58, 168.42], [560.84, 168.44], [559.32, 167.03], [539.52, 167.01]]}]\n```" - } - ] - }, - { - "id": 113, - "data_source": "Mono3DRefer", - "scene_name": "000558", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[94.45, 244.99], [161.02, 244.94], [42.96, 260.98], [-41.64, 261.06], [94.45, 181.18], [161.02, 181.19], [42.96, 180.1], [-41.64, 180.1]]}, {\"category\": \"car\", \"corners_3d\": [[575.2, 206.82], [609.8, 206.8], [611.73, 209.33], [573.07, 209.36], [575.2, 175.36], [609.8, 175.37], [611.73, 174.22], [573.07, 174.2]]}, {\"category\": \"car\", \"corners_3d\": [[218.01, 298.44], [336.63, 298.64], [148.1, 375.47], [-50.07, 374.9], [218.01, 194.58], [336.63, 194.59], [148.1, 200.94], [-50.07, 200.9]]}, {\"category\": \"car\", \"corners_3d\": [[222.31, 246.15], [291.14, 246.15], [193.94, 264.86], [103.99, 264.85], [222.31, 188.07], [291.14, 188.07], [193.94, 188.95], [103.99, 188.95]]}, {\"category\": \"car\", \"corners_3d\": [[408.41, 238.8], [467.69, 238.8], [433.66, 251.82], [359.98, 251.82], [408.41, 187.82], [467.69, 187.82], [433.66, 188.45], [359.98, 188.45]]}, {\"category\": \"car\", \"corners_3d\": [[345.96, 221.03], [388.34, 221.01], [354.9, 226.65], [305.82, 226.67], [345.96, 182.51], [388.34, 182.51], [354.9, 182.09], [305.82, 182.08]]}, {\"category\": \"car\", \"corners_3d\": [[444.42, 219.49], [487.76, 219.47], [468.65, 225.28], [417.94, 225.3], [444.42, 180.17], [487.76, 180.17], [468.65, 179.32], [417.94, 179.32]]}, {\"category\": \"van\", \"corners_3d\": [[504.42, 204.26], [540.39, 204.25], [532.39, 206.78], [491.63, 206.79], [504.42, 161.05], [540.39, 161.06], [532.39, 157.85], [491.63, 157.84]]}, {\"category\": \"van\", \"corners_3d\": [[542.58, 191.16], [560.84, 191.16], [559.32, 191.65], [539.52, 191.66], [542.58, 168.42], [560.84, 168.44], [559.32, 167.03], [539.52, 167.01]]}]\n```" - } - ] - }, - { - "id": 114, - "data_source": "Mono3DRefer", - "scene_name": "000564", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[433.44, 372.08], [248.78, 371.61], [387.48, 296.58], [502.54, 296.76], [433.44, 206.02], [248.78, 205.95], [387.48, 193.45], [502.54, 193.48]]}, {\"category\": \"car\", \"corners_3d\": [[596.66, 225.54], [545.9, 225.18], [564.17, 218.12], [608.15, 218.4], [596.66, 181.21], [545.9, 181.15], [564.17, 180.04], [608.15, 180.08]]}]\n```" - } - ] - }, - { - "id": 115, - "data_source": "Mono3DRefer", - "scene_name": "000564", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[892.64, 210.15], [881.35, 210.25], [873.9, 209.54], [884.99, 209.44], [892.64, 165.07], [881.35, 165.05], [873.9, 165.2], [884.99, 165.22]]}]\n```" - } - ] - }, - { - "id": 116, - "data_source": "Mono3DRefer", - "scene_name": "000564", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[433.44, 372.08], [248.78, 371.61], [387.48, 296.58], [502.54, 296.76], [433.44, 206.02], [248.78, 205.95], [387.48, 193.45], [502.54, 193.48]]}, {\"category\": \"car\", \"corners_3d\": [[596.66, 225.54], [545.9, 225.18], [564.17, 218.12], [608.15, 218.4], [596.66, 181.21], [545.9, 181.15], [564.17, 180.04], [608.15, 180.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[892.64, 210.15], [881.35, 210.25], [873.9, 209.54], [884.99, 209.44], [892.64, 165.07], [881.35, 165.05], [873.9, 165.2], [884.99, 165.22]]}]\n```" - } - ] - }, - { - "id": 117, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[315.91, 211.54], [284.11, 211.59], [313.17, 207.88], [341.89, 207.84], [315.91, 176.91], [284.11, 176.92], [313.17, 176.53], [341.89, 176.52]]}]\n```" - } - ] - }, - { - "id": 118, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[311.54, 218.67], [274.19, 218.67], [306.72, 214.22], [340.44, 214.22], [311.54, 186.42], [274.19, 186.42], [306.72, 185.11], [340.44, 185.1]]}, {\"category\": \"car\", \"corners_3d\": [[588.6, 194.18], [609.57, 194.18], [609.93, 195.33], [587.83, 195.33], [588.6, 174.06], [609.57, 174.06], [609.93, 174.13], [587.83, 174.13]]}]\n```" - } - ] - }, - { - "id": 119, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[617.35, 187.72], [640.18, 187.71], [649.21, 190.3], [622.39, 190.32], [617.35, 162.93], [640.18, 162.94], [649.21, 161.21], [622.39, 161.2]]}]\n```" - } - ] - }, - { - "id": 120, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[311.54, 218.67], [274.19, 218.67], [306.72, 214.22], [340.44, 214.22], [311.54, 186.42], [274.19, 186.42], [306.72, 185.11], [340.44, 185.1]]}, {\"category\": \"van\", \"corners_3d\": [[315.91, 211.54], [284.11, 211.59], [313.17, 207.88], [341.89, 207.84], [315.91, 176.91], [284.11, 176.92], [313.17, 176.53], [341.89, 176.52]]}, {\"category\": \"car\", \"corners_3d\": [[588.6, 194.18], [609.57, 194.18], [609.93, 195.33], [587.83, 195.33], [588.6, 174.06], [609.57, 174.06], [609.93, 174.13], [587.83, 174.13]]}]\n```" - } - ] - }, - { - "id": 121, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[617.35, 187.72], [640.18, 187.71], [649.21, 190.3], [622.39, 190.32], [617.35, 162.93], [640.18, 162.94], [649.21, 161.21], [622.39, 161.2]]}, {\"category\": \"van\", \"corners_3d\": [[315.91, 211.54], [284.11, 211.59], [313.17, 207.88], [341.89, 207.84], [315.91, 176.91], [284.11, 176.92], [313.17, 176.53], [341.89, 176.52]]}]\n```" - } - ] - }, - { - "id": 122, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[617.35, 187.72], [640.18, 187.71], [649.21, 190.3], [622.39, 190.32], [617.35, 162.93], [640.18, 162.94], [649.21, 161.21], [622.39, 161.2]]}, {\"category\": \"car\", \"corners_3d\": [[311.54, 218.67], [274.19, 218.67], [306.72, 214.22], [340.44, 214.22], [311.54, 186.42], [274.19, 186.42], [306.72, 185.11], [340.44, 185.1]]}, {\"category\": \"car\", \"corners_3d\": [[588.6, 194.18], [609.57, 194.18], [609.93, 195.33], [587.83, 195.33], [588.6, 174.06], [609.57, 174.06], [609.93, 174.13], [587.83, 174.13]]}]\n```" - } - ] - }, - { - "id": 123, - "data_source": "Mono3DRefer", - "scene_name": "000601", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car, bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[617.35, 187.72], [640.18, 187.71], [649.21, 190.3], [622.39, 190.32], [617.35, 162.93], [640.18, 162.94], [649.21, 161.21], [622.39, 161.2]]}, {\"category\": \"car\", \"corners_3d\": [[311.54, 218.67], [274.19, 218.67], [306.72, 214.22], [340.44, 214.22], [311.54, 186.42], [274.19, 186.42], [306.72, 185.11], [340.44, 185.1]]}, {\"category\": \"van\", \"corners_3d\": [[315.91, 211.54], [284.11, 211.59], [313.17, 207.88], [341.89, 207.84], [315.91, 176.91], [284.11, 176.92], [313.17, 176.53], [341.89, 176.52]]}, {\"category\": \"car\", \"corners_3d\": [[588.6, 194.18], [609.57, 194.18], [609.93, 195.33], [587.83, 195.33], [588.6, 174.06], [609.57, 174.06], [609.93, 174.13], [587.83, 174.13]]}]\n```" - } - ] - }, - { - "id": 124, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[681.1, 187.28], [704.47, 187.19], [745.76, 189.82], [718.38, 189.94], [681.1, 160.92], [704.47, 160.99], [745.76, 158.82], [718.38, 158.72]]}]\n```" - } - ] - }, - { - "id": 125, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[594.31, 200.59], [624.5, 200.59], [625.64, 202.8], [593.05, 202.8], [594.31, 171.63], [624.5, 171.63], [625.64, 171.53], [593.05, 171.53]]}]\n```" - } - ] - }, - { - "id": 126, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[397.69, 211.12], [346.97, 211.18], [392.71, 204.1], [434.01, 204.06], [397.69, 146.09], [346.97, 146.06], [392.71, 151.01], [434.01, 151.03]]}]\n```" - } - ] - }, - { - "id": 127, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[594.31, 200.59], [624.5, 200.59], [625.64, 202.8], [593.05, 202.8], [594.31, 171.63], [624.5, 171.63], [625.64, 171.53], [593.05, 171.53]]}, {\"category\": \"bus\", \"corners_3d\": [[681.1, 187.28], [704.47, 187.19], [745.76, 189.82], [718.38, 189.94], [681.1, 160.92], [704.47, 160.99], [745.76, 158.82], [718.38, 158.72]]}]\n```" - } - ] - }, - { - "id": 128, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[681.1, 187.28], [704.47, 187.19], [745.76, 189.82], [718.38, 189.94], [681.1, 160.92], [704.47, 160.99], [745.76, 158.82], [718.38, 158.72]]}, {\"category\": \"truck\", \"corners_3d\": [[397.69, 211.12], [346.97, 211.18], [392.71, 204.1], [434.01, 204.06], [397.69, 146.09], [346.97, 146.06], [392.71, 151.01], [434.01, 151.03]]}]\n```" - } - ] - }, - { - "id": 129, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[594.31, 200.59], [624.5, 200.59], [625.64, 202.8], [593.05, 202.8], [594.31, 171.63], [624.5, 171.63], [625.64, 171.53], [593.05, 171.53]]}, {\"category\": \"truck\", \"corners_3d\": [[397.69, 211.12], [346.97, 211.18], [392.71, 204.1], [434.01, 204.06], [397.69, 146.09], [346.97, 146.06], [392.71, 151.01], [434.01, 151.03]]}]\n```" - } - ] - }, - { - "id": 130, - "data_source": "Mono3DRefer", - "scene_name": "000630", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[594.31, 200.59], [624.5, 200.59], [625.64, 202.8], [593.05, 202.8], [594.31, 171.63], [624.5, 171.63], [625.64, 171.53], [593.05, 171.53]]}, {\"category\": \"bus\", \"corners_3d\": [[681.1, 187.28], [704.47, 187.19], [745.76, 189.82], [718.38, 189.94], [681.1, 160.92], [704.47, 160.99], [745.76, 158.82], [718.38, 158.72]]}, {\"category\": \"truck\", \"corners_3d\": [[397.69, 211.12], [346.97, 211.18], [392.71, 204.1], [434.01, 204.06], [397.69, 146.09], [346.97, 146.06], [392.71, 151.01], [434.01, 151.03]]}]\n```" - } - ] - }, - { - "id": 131, - "data_source": "Mono3DRefer", - "scene_name": "000644", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[789.61, 199.75], [781.86, 199.85], [776.65, 199.48], [784.32, 199.39], [789.61, 167.47], [781.86, 167.45], [776.65, 167.52], [784.32, 167.54]]}]\n```" - } - ] - }, - { - "id": 132, - "data_source": "Mono3DRefer", - "scene_name": "000644", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[568.28, 214.73], [526.55, 214.66], [539.09, 209.63], [575.8, 209.68], [568.28, 177.24], [526.55, 177.23], [539.09, 176.71], [575.8, 176.71]]}, {\"category\": \"car\", \"corners_3d\": [[603.11, 198.44], [575.69, 198.45], [577.71, 196.47], [603.01, 196.46], [603.11, 174.83], [575.69, 174.83], [577.71, 174.68], [603.01, 174.68]]}]\n```" - } - ] - }, - { - "id": 133, - "data_source": "Mono3DRefer", - "scene_name": "000644", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[568.28, 214.73], [526.55, 214.66], [539.09, 209.63], [575.8, 209.68], [568.28, 177.24], [526.55, 177.23], [539.09, 176.71], [575.8, 176.71]]}, {\"category\": \"car\", \"corners_3d\": [[603.11, 198.44], [575.69, 198.45], [577.71, 196.47], [603.01, 196.46], [603.11, 174.83], [575.69, 174.83], [577.71, 174.68], [603.01, 174.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[789.61, 199.75], [781.86, 199.85], [776.65, 199.48], [784.32, 199.39], [789.61, 167.47], [781.86, 167.45], [776.65, 167.52], [784.32, 167.54]]}]\n```" - } - ] - }, - { - "id": 134, - "data_source": "Mono3DRefer", - "scene_name": "000658", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[444.34, 219.71], [401.75, 219.74], [424.59, 214.39], [462.31, 214.36], [444.34, 181.39], [401.75, 181.39], [424.59, 180.42], [462.31, 180.42]]}, {\"category\": \"car\", \"corners_3d\": [[581.68, 190.05], [598.57, 190.04], [599.12, 190.98], [581.31, 190.99], [581.68, 176.75], [598.57, 176.75], [599.12, 176.97], [581.31, 176.97]]}]\n```" - } - ] - }, - { - "id": 135, - "data_source": "Mono3DRefer", - "scene_name": "000670", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[569.65, 216.77], [532.59, 216.47], [550.16, 211.98], [583.49, 212.22], [569.65, 183.32], [532.59, 183.25], [550.16, 182.18], [583.49, 182.24]]}]\n```" - } - ] - }, - { - "id": 136, - "data_source": "Mono3DRefer", - "scene_name": "000729", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[517.28, 218.51], [509.93, 221.88], [345.84, 221.77], [364.44, 218.42], [517.28, 143.02], [509.93, 140.81], [345.84, 140.88], [364.44, 143.08]]}, {\"category\": \"van\", \"corners_3d\": [[782.26, 316.2], [955.25, 316.23], [2334.09, 889.05], [1469.31, 888.36], [782.26, 92.88], [955.25, 92.86], [2334.09, -226.72], [1469.31, -226.34]]}]\n```" - } - ] - }, - { - "id": 137, - "data_source": "Mono3DRefer", - "scene_name": "000729", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[504.32, 218.85], [499.37, 222.14], [359.87, 222.62], [374.29, 219.26], [504.32, 175.17], [499.37, 175.34], [359.87, 175.36], [374.29, 175.19]]}, {\"category\": \"car\", \"corners_3d\": [[665.86, 219.97], [713.58, 219.97], [734.92, 229.69], [677.36, 229.69], [665.86, 178.44], [713.58, 178.44], [734.92, 179.59], [677.36, 179.59]]}, {\"category\": \"car\", \"corners_3d\": [[813.87, 210.13], [823.08, 212.36], [734.76, 212.03], [730.57, 209.83], [813.87, 173.62], [823.08, 173.66], [734.76, 173.65], [730.57, 173.61]]}, {\"category\": \"car\", \"corners_3d\": [[185.18, 327.7], [55.65, 374.17], [-452.17, 372.74], [-206.77, 326.85], [185.18, 164.05], [55.65, 161.41], [-452.17, 161.49], [-206.77, 164.1]]}, {\"category\": \"car\", \"corners_3d\": [[-160.27, 317.45], [-31.13, 291.96], [299.29, 289.26], [242.46, 313.5], [-160.27, 178.28], [-31.13, 177.32], [299.29, 177.22], [242.46, 178.13]]}, {\"category\": \"car\", \"corners_3d\": [[428.66, 245.83], [403.93, 255.42], [181.49, 255.17], [231.92, 245.63], [428.66, 173.31], [403.93, 173.37], [181.49, 173.37], [231.92, 173.31]]}]\n```" - } - ] - }, - { - "id": 138, - "data_source": "Mono3DRefer", - "scene_name": "000729", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[504.32, 218.85], [499.37, 222.14], [359.87, 222.62], [374.29, 219.26], [504.32, 175.17], [499.37, 175.34], [359.87, 175.36], [374.29, 175.19]]}, {\"category\": \"van\", \"corners_3d\": [[517.28, 218.51], [509.93, 221.88], [345.84, 221.77], [364.44, 218.42], [517.28, 143.02], [509.93, 140.81], [345.84, 140.88], [364.44, 143.08]]}, {\"category\": \"car\", \"corners_3d\": [[665.86, 219.97], [713.58, 219.97], [734.92, 229.69], [677.36, 229.69], [665.86, 178.44], [713.58, 178.44], [734.92, 179.59], [677.36, 179.59]]}, {\"category\": \"car\", \"corners_3d\": [[813.87, 210.13], [823.08, 212.36], [734.76, 212.03], [730.57, 209.83], [813.87, 173.62], [823.08, 173.66], [734.76, 173.65], [730.57, 173.61]]}, {\"category\": \"van\", \"corners_3d\": [[782.26, 316.2], [955.25, 316.23], [2334.09, 889.05], [1469.31, 888.36], [782.26, 92.88], [955.25, 92.86], [2334.09, -226.72], [1469.31, -226.34]]}, {\"category\": \"car\", \"corners_3d\": [[185.18, 327.7], [55.65, 374.17], [-452.17, 372.74], [-206.77, 326.85], [185.18, 164.05], [55.65, 161.41], [-452.17, 161.49], [-206.77, 164.1]]}, {\"category\": \"car\", \"corners_3d\": [[-160.27, 317.45], [-31.13, 291.96], [299.29, 289.26], [242.46, 313.5], [-160.27, 178.28], [-31.13, 177.32], [299.29, 177.22], [242.46, 178.13]]}, {\"category\": \"car\", \"corners_3d\": [[428.66, 245.83], [403.93, 255.42], [181.49, 255.17], [231.92, 245.63], [428.66, 173.31], [403.93, 173.37], [181.49, 173.37], [231.92, 173.31]]}]\n```" - } - ] - }, - { - "id": 139, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[651.34, 220.4], [694.25, 220.29], [714.48, 228.79], [663.92, 228.94], [651.34, 175.18], [694.25, 175.18], [714.48, 175.59], [663.92, 175.6]]}, {\"category\": \"van\", \"corners_3d\": [[640.91, 207.04], [670.77, 207.0], [680.79, 211.16], [647.3, 211.21], [640.91, 173.02], [670.77, 173.02], [680.79, 173.04], [647.3, 173.04]]}]\n```" - } - ] - }, - { - "id": 140, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[809.22, 252.14], [826.69, 252.12], [847.46, 259.48], [828.38, 259.5], [809.22, 182.08], [826.69, 182.08], [847.46, 182.93], [828.38, 182.94]]}]\n```" - } - ] - }, - { - "id": 141, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[509.31, 266.86], [425.11, 267.2], [456.7, 248.82], [524.46, 248.61], [509.31, 190.6], [425.11, 190.66], [456.7, 187.19], [524.46, 187.15]]}, {\"category\": \"car\", \"corners_3d\": [[677.63, 243.19], [746.25, 243.13], [781.68, 260.5], [696.12, 260.6], [677.63, 176.88], [746.25, 176.88], [781.68, 177.87], [696.12, 177.88]]}, {\"category\": \"car\", \"corners_3d\": [[665.31, 230.14], [714.32, 230.07], [735.71, 240.38], [677.88, 240.48], [665.31, 185.54], [714.32, 185.52], [735.71, 187.81], [677.88, 187.83]]}, {\"category\": \"car\", \"corners_3d\": [[566.36, 212.31], [534.3, 212.32], [541.36, 208.2], [570.06, 208.18], [566.36, 182.76], [534.3, 182.76], [541.36, 181.73], [570.06, 181.72]]}]\n```" - } - ] - }, - { - "id": 142, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[651.34, 220.4], [694.25, 220.29], [714.48, 228.79], [663.92, 228.94], [651.34, 175.18], [694.25, 175.18], [714.48, 175.59], [663.92, 175.6]]}, {\"category\": \"van\", \"corners_3d\": [[640.91, 207.04], [670.77, 207.0], [680.79, 211.16], [647.3, 211.21], [640.91, 173.02], [670.77, 173.02], [680.79, 173.04], [647.3, 173.04]]}, {\"category\": \"cyclist\", \"corners_3d\": [[809.22, 252.14], [826.69, 252.12], [847.46, 259.48], [828.38, 259.5], [809.22, 182.08], [826.69, 182.08], [847.46, 182.93], [828.38, 182.94]]}]\n```" - } - ] - }, - { - "id": 143, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[509.31, 266.86], [425.11, 267.2], [456.7, 248.82], [524.46, 248.61], [509.31, 190.6], [425.11, 190.66], [456.7, 187.19], [524.46, 187.15]]}, {\"category\": \"car\", \"corners_3d\": [[677.63, 243.19], [746.25, 243.13], [781.68, 260.5], [696.12, 260.6], [677.63, 176.88], [746.25, 176.88], [781.68, 177.87], [696.12, 177.88]]}, {\"category\": \"car\", \"corners_3d\": [[665.31, 230.14], [714.32, 230.07], [735.71, 240.38], [677.88, 240.48], [665.31, 185.54], [714.32, 185.52], [735.71, 187.81], [677.88, 187.83]]}, {\"category\": \"van\", \"corners_3d\": [[651.34, 220.4], [694.25, 220.29], [714.48, 228.79], [663.92, 228.94], [651.34, 175.18], [694.25, 175.18], [714.48, 175.59], [663.92, 175.6]]}, {\"category\": \"car\", \"corners_3d\": [[566.36, 212.31], [534.3, 212.32], [541.36, 208.2], [570.06, 208.18], [566.36, 182.76], [534.3, 182.76], [541.36, 181.73], [570.06, 181.72]]}, {\"category\": \"van\", \"corners_3d\": [[640.91, 207.04], [670.77, 207.0], [680.79, 211.16], [647.3, 211.21], [640.91, 173.02], [670.77, 173.02], [680.79, 173.04], [647.3, 173.04]]}]\n```" - } - ] - }, - { - "id": 144, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[509.31, 266.86], [425.11, 267.2], [456.7, 248.82], [524.46, 248.61], [509.31, 190.6], [425.11, 190.66], [456.7, 187.19], [524.46, 187.15]]}, {\"category\": \"car\", \"corners_3d\": [[677.63, 243.19], [746.25, 243.13], [781.68, 260.5], [696.12, 260.6], [677.63, 176.88], [746.25, 176.88], [781.68, 177.87], [696.12, 177.88]]}, {\"category\": \"car\", \"corners_3d\": [[665.31, 230.14], [714.32, 230.07], [735.71, 240.38], [677.88, 240.48], [665.31, 185.54], [714.32, 185.52], [735.71, 187.81], [677.88, 187.83]]}, {\"category\": \"car\", \"corners_3d\": [[566.36, 212.31], [534.3, 212.32], [541.36, 208.2], [570.06, 208.18], [566.36, 182.76], [534.3, 182.76], [541.36, 181.73], [570.06, 181.72]]}, {\"category\": \"cyclist\", \"corners_3d\": [[809.22, 252.14], [826.69, 252.12], [847.46, 259.48], [828.38, 259.5], [809.22, 182.08], [826.69, 182.08], [847.46, 182.93], [828.38, 182.94]]}]\n```" - } - ] - }, - { - "id": 145, - "data_source": "Mono3DRefer", - "scene_name": "000754", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[509.31, 266.86], [425.11, 267.2], [456.7, 248.82], [524.46, 248.61], [509.31, 190.6], [425.11, 190.66], [456.7, 187.19], [524.46, 187.15]]}, {\"category\": \"car\", \"corners_3d\": [[677.63, 243.19], [746.25, 243.13], [781.68, 260.5], [696.12, 260.6], [677.63, 176.88], [746.25, 176.88], [781.68, 177.87], [696.12, 177.88]]}, {\"category\": \"car\", \"corners_3d\": [[665.31, 230.14], [714.32, 230.07], [735.71, 240.38], [677.88, 240.48], [665.31, 185.54], [714.32, 185.52], [735.71, 187.81], [677.88, 187.83]]}, {\"category\": \"van\", \"corners_3d\": [[651.34, 220.4], [694.25, 220.29], [714.48, 228.79], [663.92, 228.94], [651.34, 175.18], [694.25, 175.18], [714.48, 175.59], [663.92, 175.6]]}, {\"category\": \"car\", \"corners_3d\": [[566.36, 212.31], [534.3, 212.32], [541.36, 208.2], [570.06, 208.18], [566.36, 182.76], [534.3, 182.76], [541.36, 181.73], [570.06, 181.72]]}, {\"category\": \"van\", \"corners_3d\": [[640.91, 207.04], [670.77, 207.0], [680.79, 211.16], [647.3, 211.21], [640.91, 173.02], [670.77, 173.02], [680.79, 173.04], [647.3, 173.04]]}, {\"category\": \"cyclist\", \"corners_3d\": [[809.22, 252.14], [826.69, 252.12], [847.46, 259.48], [828.38, 259.5], [809.22, 182.08], [826.69, 182.08], [847.46, 182.93], [828.38, 182.94]]}]\n```" - } - ] - }, - { - "id": 146, - "data_source": "Mono3DRefer", - "scene_name": "000766", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[823.72, 351.38], [1007.55, 350.5], [1704.05, 650.94], [1214.51, 657.35], [823.72, 136.48], [1007.55, 136.66], [1704.05, 75.45], [1214.51, 74.14]]}]\n```" - } - ] - }, - { - "id": 147, - "data_source": "Mono3DRefer", - "scene_name": "000766", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[844.04, 210.04], [853.65, 211.84], [769.09, 211.61], [763.42, 209.83], [844.04, 175.8], [853.65, 175.95], [769.09, 175.93], [763.42, 175.79]]}, {\"category\": \"car\", \"corners_3d\": [[661.36, 192.73], [680.69, 192.76], [683.01, 194.1], [662.37, 194.07], [661.36, 174.46], [680.69, 174.47], [683.01, 174.58], [662.37, 174.57]]}]\n```" - } - ] - }, - { - "id": 148, - "data_source": "Mono3DRefer", - "scene_name": "000766", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[823.72, 351.38], [1007.55, 350.5], [1704.05, 650.94], [1214.51, 657.35], [823.72, 136.48], [1007.55, 136.66], [1704.05, 75.45], [1214.51, 74.14]]}, {\"category\": \"car\", \"corners_3d\": [[844.04, 210.04], [853.65, 211.84], [769.09, 211.61], [763.42, 209.83], [844.04, 175.8], [853.65, 175.95], [769.09, 175.93], [763.42, 175.79]]}, {\"category\": \"car\", \"corners_3d\": [[661.36, 192.73], [680.69, 192.76], [683.01, 194.1], [662.37, 194.07], [661.36, 174.46], [680.69, 174.47], [683.01, 174.58], [662.37, 174.57]]}]\n```" - } - ] - }, - { - "id": 149, - "data_source": "Mono3DRefer", - "scene_name": "000772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[471.71, 259.54], [451.13, 251.13], [631.87, 242.54], [667.07, 249.13], [471.71, 155.18], [451.13, 156.9], [631.87, 158.65], [667.07, 157.31]]}]\n```" - } - ] - }, - { - "id": 150, - "data_source": "Mono3DRefer", - "scene_name": "000772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1228.41, 252.78], [1338.5, 261.23], [1260.08, 275.02], [1143.22, 263.89], [1228.41, 173.79], [1338.5, 173.89], [1260.08, 174.06], [1143.22, 173.92]]}, {\"category\": \"car\", \"corners_3d\": [[947.39, 233.9], [901.87, 229.91], [990.04, 225.57], [1037.54, 228.95], [947.39, 176.11], [901.87, 175.9], [990.04, 175.67], [1037.54, 175.85]]}, {\"category\": \"car\", \"corners_3d\": [[1157.24, 223.11], [1108.13, 220.39], [1172.7, 217.16], [1221.74, 219.52], [1157.24, 179.9], [1108.13, 179.52], [1172.7, 179.06], [1221.74, 179.4]]}]\n```" - } - ] - }, - { - "id": 151, - "data_source": "Mono3DRefer", - "scene_name": "000772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1228.41, 252.78], [1338.5, 261.23], [1260.08, 275.02], [1143.22, 263.89], [1228.41, 173.79], [1338.5, 173.89], [1260.08, 174.06], [1143.22, 173.92]]}, {\"category\": \"van\", \"corners_3d\": [[471.71, 259.54], [451.13, 251.13], [631.87, 242.54], [667.07, 249.13], [471.71, 155.18], [451.13, 156.9], [631.87, 158.65], [667.07, 157.31]]}, {\"category\": \"car\", \"corners_3d\": [[947.39, 233.9], [901.87, 229.91], [990.04, 225.57], [1037.54, 228.95], [947.39, 176.11], [901.87, 175.9], [990.04, 175.67], [1037.54, 175.85]]}, {\"category\": \"car\", \"corners_3d\": [[1157.24, 223.11], [1108.13, 220.39], [1172.7, 217.16], [1221.74, 219.52], [1157.24, 179.9], [1108.13, 179.52], [1172.7, 179.06], [1221.74, 179.4]]}]\n```" - } - ] - }, - { - "id": 152, - "data_source": "Mono3DRefer", - "scene_name": "000790", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[711.93, 224.08], [658.5, 224.71], [631.8, 216.37], [677.0, 215.92], [711.93, 167.11], [658.5, 167.04], [631.8, 167.98], [677.0, 168.03]]}, {\"category\": \"car\", \"corners_3d\": [[461.21, 231.65], [413.02, 232.19], [427.22, 223.26], [468.1, 222.87], [461.21, 186.05], [413.02, 186.17], [427.22, 184.17], [468.1, 184.08]]}, {\"category\": \"car\", \"corners_3d\": [[481.88, 217.31], [445.88, 217.58], [453.64, 213.02], [485.95, 212.8], [481.88, 188.2], [445.88, 188.29], [453.64, 186.72], [485.95, 186.65]]}]\n```" - } - ] - }, - { - "id": 153, - "data_source": "Mono3DRefer", - "scene_name": "000795", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[980.89, 231.19], [992.6, 227.6], [1200.64, 236.3], [1202.88, 241.17], [980.89, 172.84], [992.6, 172.84], [1200.64, 172.84], [1202.88, 172.84]]}, {\"category\": \"car\", \"corners_3d\": [[541.17, 219.98], [501.86, 219.75], [521.43, 214.42], [556.35, 214.6], [541.17, 182.96], [501.86, 182.91], [521.43, 181.77], [556.35, 181.81]]}, {\"category\": \"car\", \"corners_3d\": [[1098.25, 246.46], [1107.57, 240.15], [1362.28, 252.67], [1379.84, 261.7], [1098.25, 170.89], [1107.57, 171.05], [1362.28, 170.72], [1379.84, 170.48]]}, {\"category\": \"car\", \"corners_3d\": [[690.9, 200.35], [716.79, 200.47], [719.03, 203.61], [690.19, 203.46], [690.9, 174.73], [716.79, 174.74], [719.03, 174.95], [690.19, 174.94]]}, {\"category\": \"car\", \"corners_3d\": [[349.52, 218.16], [335.82, 220.77], [240.58, 220.96], [259.52, 218.33], [349.52, 186.69], [335.82, 187.49], [240.58, 187.55], [259.52, 186.74]]}, {\"category\": \"car\", \"corners_3d\": [[753.59, 256.42], [839.18, 256.82], [913.21, 287.88], [795.69, 287.13], [753.59, 188.97], [839.18, 189.04], [913.21, 195.03], [795.69, 194.89]]}, {\"category\": \"car\", \"corners_3d\": [[1207.04, 261.76], [1200.27, 253.12], [1535.91, 270.2], [1589.16, 283.21], [1207.04, 155.78], [1200.27, 157.44], [1535.91, 154.16], [1589.16, 151.66]]}, {\"category\": \"car\", \"corners_3d\": [[722.4, 234.82], [779.95, 235.02], [815.43, 250.67], [743.31, 250.35], [722.4, 182.0], [779.95, 182.03], [815.43, 184.34], [743.31, 184.29]]}, {\"category\": \"car\", \"corners_3d\": [[519.07, 227.57], [474.75, 227.2], [500.92, 220.55], [539.94, 220.84], [519.07, 184.76], [474.75, 184.68], [500.92, 183.24], [539.94, 183.3]]}]\n```" - } - ] - }, - { - "id": 154, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[118.98, 517.63], [-366.19, 535.43], [254.25, 290.2], [406.47, 288.27], [118.98, -145.54], [-366.19, -161.97], [254.25, 64.49], [406.47, 66.27]]}]\n```" - } - ] - }, - { - "id": 155, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[605.6, 264.43], [629.14, 262.22], [644.74, 264.73], [620.92, 267.07], [605.6, 163.85], [629.14, 164.07], [644.74, 163.82], [620.92, 163.59]]}]\n```" - } - ] - }, - { - "id": 156, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[627.22, 252.53], [708.95, 252.44], [741.47, 276.85], [634.69, 277.0], [627.22, 174.44], [708.95, 174.44], [741.47, 174.92], [634.69, 174.93]]}]\n```" - } - ] - }, - { - "id": 157, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[118.98, 517.63], [-366.19, 535.43], [254.25, 290.2], [406.47, 288.27], [118.98, -145.54], [-366.19, -161.97], [254.25, 64.49], [406.47, 66.27]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[605.6, 264.43], [629.14, 262.22], [644.74, 264.73], [620.92, 267.07], [605.6, 163.85], [629.14, 164.07], [644.74, 163.82], [620.92, 163.59]]}]\n```" - } - ] - }, - { - "id": 158, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[118.98, 517.63], [-366.19, 535.43], [254.25, 290.2], [406.47, 288.27], [118.98, -145.54], [-366.19, -161.97], [254.25, 64.49], [406.47, 66.27]]}, {\"category\": \"car\", \"corners_3d\": [[627.22, 252.53], [708.95, 252.44], [741.47, 276.85], [634.69, 277.0], [627.22, 174.44], [708.95, 174.44], [741.47, 174.92], [634.69, 174.93]]}]\n```" - } - ] - }, - { - "id": 159, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[627.22, 252.53], [708.95, 252.44], [741.47, 276.85], [634.69, 277.0], [627.22, 174.44], [708.95, 174.44], [741.47, 174.92], [634.69, 174.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[605.6, 264.43], [629.14, 262.22], [644.74, 264.73], [620.92, 267.07], [605.6, 163.85], [629.14, 164.07], [644.74, 163.82], [620.92, 163.59]]}]\n```" - } - ] - }, - { - "id": 160, - "data_source": "Mono3DRefer", - "scene_name": "000801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[118.98, 517.63], [-366.19, 535.43], [254.25, 290.2], [406.47, 288.27], [118.98, -145.54], [-366.19, -161.97], [254.25, 64.49], [406.47, 66.27]]}, {\"category\": \"car\", \"corners_3d\": [[627.22, 252.53], [708.95, 252.44], [741.47, 276.85], [634.69, 277.0], [627.22, 174.44], [708.95, 174.44], [741.47, 174.92], [634.69, 174.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[605.6, 264.43], [629.14, 262.22], [644.74, 264.73], [620.92, 267.07], [605.6, 163.85], [629.14, 164.07], [644.74, 163.82], [620.92, 163.59]]}]\n```" - } - ] - }, - { - "id": 161, - "data_source": "Mono3DRefer", - "scene_name": "000806", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[484.75, 205.17], [470.57, 205.16], [472.59, 204.74], [486.58, 204.75], [484.75, 177.45], [470.57, 177.44], [472.59, 177.38], [486.58, 177.39]]}]\n```" - } - ] - }, - { - "id": 162, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[590.74, 214.9], [635.1, 214.9], [638.14, 220.01], [588.38, 220.01], [590.74, 172.33], [635.1, 172.33], [638.14, 172.27], [588.38, 172.27]]}]\n```" - } - ] - }, - { - "id": 163, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[511.96, 205.13], [556.84, 205.13], [526.35, 223.59], [455.8, 223.59], [511.96, 142.64], [556.84, 142.64], [526.35, 125.35], [455.8, 125.35]]}]\n```" - } - ] - }, - { - "id": 164, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[793.27, 229.04], [811.86, 228.89], [842.22, 235.1], [821.66, 235.28], [793.27, 163.05], [811.86, 163.08], [842.22, 161.99], [821.66, 161.96]]}]\n```" - } - ] - }, - { - "id": 165, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[590.74, 214.9], [635.1, 214.9], [638.14, 220.01], [588.38, 220.01], [590.74, 172.33], [635.1, 172.33], [638.14, 172.27], [588.38, 172.27]]}, {\"category\": \"tram\", \"corners_3d\": [[511.96, 205.13], [556.84, 205.13], [526.35, 223.59], [455.8, 223.59], [511.96, 142.64], [556.84, 142.64], [526.35, 125.35], [455.8, 125.35]]}]\n```" - } - ] - }, - { - "id": 166, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[590.74, 214.9], [635.1, 214.9], [638.14, 220.01], [588.38, 220.01], [590.74, 172.33], [635.1, 172.33], [638.14, 172.27], [588.38, 172.27]]}, {\"category\": \"cyclist\", \"corners_3d\": [[793.27, 229.04], [811.86, 228.89], [842.22, 235.1], [821.66, 235.28], [793.27, 163.05], [811.86, 163.08], [842.22, 161.99], [821.66, 161.96]]}]\n```" - } - ] - }, - { - "id": 167, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[793.27, 229.04], [811.86, 228.89], [842.22, 235.1], [821.66, 235.28], [793.27, 163.05], [811.86, 163.08], [842.22, 161.99], [821.66, 161.96]]}, {\"category\": \"tram\", \"corners_3d\": [[511.96, 205.13], [556.84, 205.13], [526.35, 223.59], [455.8, 223.59], [511.96, 142.64], [556.84, 142.64], [526.35, 125.35], [455.8, 125.35]]}]\n```" - } - ] - }, - { - "id": 168, - "data_source": "Mono3DRefer", - "scene_name": "000812", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[590.74, 214.9], [635.1, 214.9], [638.14, 220.01], [588.38, 220.01], [590.74, 172.33], [635.1, 172.33], [638.14, 172.27], [588.38, 172.27]]}, {\"category\": \"cyclist\", \"corners_3d\": [[793.27, 229.04], [811.86, 228.89], [842.22, 235.1], [821.66, 235.28], [793.27, 163.05], [811.86, 163.08], [842.22, 161.99], [821.66, 161.96]]}, {\"category\": \"tram\", \"corners_3d\": [[511.96, 205.13], [556.84, 205.13], [526.35, 223.59], [455.8, 223.59], [511.96, 142.64], [556.84, 142.64], [526.35, 125.35], [455.8, 125.35]]}]\n```" - } - ] - }, - { - "id": 169, - "data_source": "Mono3DRefer", - "scene_name": "000838", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[739.61, 209.71], [747.24, 211.4], [659.98, 211.64], [656.16, 209.92], [739.61, 180.97], [747.24, 181.35], [659.98, 181.4], [656.16, 181.02]]}]\n```" - } - ] - }, - { - "id": 170, - "data_source": "Mono3DRefer", - "scene_name": "000843", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-49.87, 412.43], [-247.61, 413.13], [76.41, 321.47], [198.26, 321.2], [-49.87, 232.4], [-247.61, 232.57], [76.41, 209.79], [198.26, 209.72]]}, {\"category\": \"car\", \"corners_3d\": [[735.15, 261.23], [834.34, 261.24], [929.01, 298.56], [787.93, 298.54], [735.15, 175.67], [834.34, 175.67], [929.01, 176.86], [787.93, 176.85]]}, {\"category\": \"car\", \"corners_3d\": [[272.17, 293.89], [174.44, 293.74], [291.89, 261.6], [363.72, 261.68], [272.17, 205.37], [174.44, 205.33], [291.89, 196.7], [363.72, 196.72]]}, {\"category\": \"car\", \"corners_3d\": [[421.19, 245.67], [359.54, 245.68], [403.3, 232.9], [454.13, 232.9], [421.19, 180.68], [359.54, 180.68], [403.3, 179.31], [454.13, 179.31]]}, {\"category\": \"car\", \"corners_3d\": [[575.83, 199.48], [549.97, 199.46], [556.02, 197.27], [579.74, 197.29], [575.83, 178.73], [549.97, 178.73], [556.02, 178.25], [579.74, 178.25]]}, {\"category\": \"car\", \"corners_3d\": [[462.76, 225.88], [417.22, 225.82], [444.82, 218.73], [484.29, 218.78], [462.76, 180.94], [417.22, 180.93], [444.82, 179.85], [484.29, 179.86]]}, {\"category\": \"car\", \"corners_3d\": [[505.87, 211.36], [472.53, 211.34], [485.81, 207.78], [516.07, 207.8], [505.87, 183.26], [472.53, 183.26], [485.81, 182.29], [516.07, 182.3]]}, {\"category\": \"car\", \"corners_3d\": [[639.91, 193.22], [659.14, 193.23], [662.12, 194.68], [641.52, 194.67], [639.91, 175.28], [659.14, 175.28], [662.12, 175.45], [641.52, 175.45]]}, {\"category\": \"car\", \"corners_3d\": [[700.01, 189.99], [697.2, 189.45], [759.47, 189.45], [764.31, 189.99], [700.01, 169.67], [697.2, 169.77], [759.47, 169.77], [764.31, 169.67]]}]\n```" - } - ] - }, - { - "id": 171, - "data_source": "Mono3DRefer", - "scene_name": "000852", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[476.0, 255.03], [407.1, 255.27], [442.12, 239.25], [497.58, 239.1], [476.0, 190.85], [407.1, 190.9], [442.12, 187.39], [497.58, 187.36]]}, {\"category\": \"car\", \"corners_3d\": [[561.81, 205.04], [529.7, 205.09], [535.29, 201.96], [564.28, 201.93], [561.81, 178.34], [529.7, 178.34], [535.29, 177.81], [564.28, 177.8]]}]\n```" - } - ] - }, - { - "id": 172, - "data_source": "Mono3DRefer", - "scene_name": "000852", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[512.66, 231.21], [456.96, 231.22], [481.05, 221.97], [527.92, 221.97], [512.66, 170.49], [456.96, 170.49], [481.05, 170.87], [527.92, 170.87]]}]\n```" - } - ] - }, - { - "id": 173, - "data_source": "Mono3DRefer", - "scene_name": "000852", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[476.0, 255.03], [407.1, 255.27], [442.12, 239.25], [497.58, 239.1], [476.0, 190.85], [407.1, 190.9], [442.12, 187.39], [497.58, 187.36]]}, {\"category\": \"van\", \"corners_3d\": [[512.66, 231.21], [456.96, 231.22], [481.05, 221.97], [527.92, 221.97], [512.66, 170.49], [456.96, 170.49], [481.05, 170.87], [527.92, 170.87]]}, {\"category\": \"car\", \"corners_3d\": [[561.81, 205.04], [529.7, 205.09], [535.29, 201.96], [564.28, 201.93], [561.81, 178.34], [529.7, 178.34], [535.29, 177.81], [564.28, 177.8]]}]\n```" - } - ] - }, - { - "id": 174, - "data_source": "Mono3DRefer", - "scene_name": "000863", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[856.27, 372.73], [1059.29, 376.62], [1400.72, 551.18], [1017.93, 537.97], [856.27, 176.56], [1059.29, 176.63], [1400.72, 179.86], [1017.93, 179.62]]}, {\"category\": \"car\", \"corners_3d\": [[-174.02, 770.88], [-603.86, 716.99], [229.51, 370.08], [404.2, 376.74], [-174.02, 230.73], [-603.86, 225.51], [229.51, 191.94], [404.2, 192.58]]}, {\"category\": \"car\", \"corners_3d\": [[663.18, 201.7], [636.99, 201.46], [647.32, 199.36], [671.65, 199.57], [663.18, 178.18], [636.99, 178.14], [647.32, 177.75], [671.65, 177.79]]}]\n```" - } - ] - }, - { - "id": 175, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[487.98, 204.39], [465.37, 204.43], [475.29, 201.7], [495.93, 201.66], [487.98, 172.26], [465.37, 172.26], [475.29, 172.31], [495.93, 172.31]]}]\n```" - } - ] - }, - { - "id": 176, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[92.33, 286.03], [-75.31, 287.8], [150.38, 246.87], [256.66, 246.13], [92.33, 93.82], [-75.31, 92.58], [150.38, 121.17], [256.66, 121.69]]}]\n```" - } - ] - }, - { - "id": 177, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[550.04, 194.52], [566.19, 194.48], [567.29, 195.6], [550.3, 195.65], [550.04, 181.82], [566.19, 181.8], [567.29, 182.27], [550.3, 182.29]]}]\n```" - } - ] - }, - { - "id": 178, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[92.33, 286.03], [-75.31, 287.8], [150.38, 246.87], [256.66, 246.13], [92.33, 93.82], [-75.31, 92.58], [150.38, 121.17], [256.66, 121.69]]}, {\"category\": \"van\", \"corners_3d\": [[487.98, 204.39], [465.37, 204.43], [475.29, 201.7], [495.93, 201.66], [487.98, 172.26], [465.37, 172.26], [475.29, 172.31], [495.93, 172.31]]}]\n```" - } - ] - }, - { - "id": 179, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[487.98, 204.39], [465.37, 204.43], [475.29, 201.7], [495.93, 201.66], [487.98, 172.26], [465.37, 172.26], [475.29, 172.31], [495.93, 172.31]]}, {\"category\": \"car\", \"corners_3d\": [[550.04, 194.52], [566.19, 194.48], [567.29, 195.6], [550.3, 195.65], [550.04, 181.82], [566.19, 181.8], [567.29, 182.27], [550.3, 182.29]]}]\n```" - } - ] - }, - { - "id": 180, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[92.33, 286.03], [-75.31, 287.8], [150.38, 246.87], [256.66, 246.13], [92.33, 93.82], [-75.31, 92.58], [150.38, 121.17], [256.66, 121.69]]}, {\"category\": \"car\", \"corners_3d\": [[550.04, 194.52], [566.19, 194.48], [567.29, 195.6], [550.3, 195.65], [550.04, 181.82], [566.19, 181.8], [567.29, 182.27], [550.3, 182.29]]}]\n```" - } - ] - }, - { - "id": 181, - "data_source": "Mono3DRefer", - "scene_name": "000873", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[92.33, 286.03], [-75.31, 287.8], [150.38, 246.87], [256.66, 246.13], [92.33, 93.82], [-75.31, 92.58], [150.38, 121.17], [256.66, 121.69]]}, {\"category\": \"van\", \"corners_3d\": [[487.98, 204.39], [465.37, 204.43], [475.29, 201.7], [495.93, 201.66], [487.98, 172.26], [465.37, 172.26], [475.29, 172.31], [495.93, 172.31]]}, {\"category\": \"car\", \"corners_3d\": [[550.04, 194.52], [566.19, 194.48], [567.29, 195.6], [550.3, 195.65], [550.04, 181.82], [566.19, 181.8], [567.29, 182.27], [550.3, 182.29]]}]\n```" - } - ] - }, - { - "id": 182, - "data_source": "Mono3DRefer", - "scene_name": "000949", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[142.56, 246.51], [73.27, 246.52], [174.97, 232.53], [231.1, 232.53], [142.56, 189.42], [73.27, 189.42], [174.97, 186.27], [231.1, 186.27]]}]\n```" - } - ] - }, - { - "id": 183, - "data_source": "Mono3DRefer", - "scene_name": "000949", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[124.33, 233.48], [47.45, 233.48], [181.66, 218.99], [240.16, 218.99], [124.33, 123.97], [47.45, 123.97], [181.66, 135.65], [240.16, 135.66]]}, {\"category\": \"truck\", \"corners_3d\": [[394.22, 201.12], [363.89, 201.13], [389.62, 198.07], [416.67, 198.06], [394.22, 160.73], [363.89, 160.73], [389.62, 162.04], [416.67, 162.05]]}, {\"category\": \"truck\", \"corners_3d\": [[-157.36, 268.75], [-280.26, 268.76], [19.18, 236.47], [100.67, 236.46], [-157.36, 97.6], [-280.26, 97.59], [19.18, 122.93], [100.67, 122.94]]}]\n```" - } - ] - }, - { - "id": 184, - "data_source": "Mono3DRefer", - "scene_name": "000949", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[124.33, 233.48], [47.45, 233.48], [181.66, 218.99], [240.16, 218.99], [124.33, 123.97], [47.45, 123.97], [181.66, 135.65], [240.16, 135.66]]}, {\"category\": \"truck\", \"corners_3d\": [[394.22, 201.12], [363.89, 201.13], [389.62, 198.07], [416.67, 198.06], [394.22, 160.73], [363.89, 160.73], [389.62, 162.04], [416.67, 162.05]]}, {\"category\": \"car\", \"corners_3d\": [[142.56, 246.51], [73.27, 246.52], [174.97, 232.53], [231.1, 232.53], [142.56, 189.42], [73.27, 189.42], [174.97, 186.27], [231.1, 186.27]]}, {\"category\": \"truck\", \"corners_3d\": [[-157.36, 268.75], [-280.26, 268.76], [19.18, 236.47], [100.67, 236.46], [-157.36, 97.6], [-280.26, 97.59], [19.18, 122.93], [100.67, 122.94]]}]\n```" - } - ] - }, - { - "id": 185, - "data_source": "Mono3DRefer", - "scene_name": "000966", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[660.4, 202.05], [667.02, 201.6], [685.66, 201.89], [679.34, 202.34], [660.4, 146.15], [667.02, 146.86], [685.66, 146.4], [679.34, 145.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[304.49, 256.71], [272.51, 260.9], [241.15, 259.15], [274.07, 255.13], [304.49, 141.34], [272.51, 139.19], [241.15, 140.09], [274.07, 142.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1471.77, 665.05], [1293.29, 682.5], [1052.97, 526.84], [1183.28, 518.45], [1471.77, 118.17], [1293.29, 115.92], [1052.97, 135.95], [1183.28, 137.03]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[895.56, 329.49], [817.81, 330.17], [787.43, 311.43], [855.6, 310.91], [895.56, 139.35], [817.81, 139.16], [787.43, 144.34], [855.6, 144.49]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[812.21, 332.01], [733.78, 335.84], [701.79, 318.16], [772.19, 315.14], [812.21, 147.75], [733.78, 146.92], [701.79, 150.75], [772.19, 151.4]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1021.45, 333.71], [991.63, 342.0], [891.29, 324.08], [922.56, 317.48], [1021.45, 140.35], [991.63, 138.17], [891.29, 142.87], [922.56, 144.6]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1124.18, 337.74], [1094.57, 343.1], [979.99, 321.19], [1009.01, 317.16], [1124.18, 149.05], [1094.57, 147.98], [979.99, 152.36], [1009.01, 153.16]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[885.45, 236.85], [885.73, 235.36], [932.07, 236.64], [933.07, 238.2], [885.45, 161.23], [885.73, 161.74], [932.07, 161.3], [933.07, 160.77]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[579.45, 319.83], [523.99, 323.65], [505.31, 307.99], [555.3, 304.94], [579.45, 144.24], [523.99, 143.25], [505.31, 147.32], [555.3, 148.12]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[608.57, 300.83], [564.24, 303.25], [546.43, 291.58], [586.94, 289.59], [608.57, 158.37], [564.24, 157.92], [546.43, 160.07], [586.94, 160.44]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[645.14, 298.21], [599.45, 300.1], [582.27, 287.94], [623.62, 286.41], [645.14, 140.22], [599.45, 139.57], [582.27, 143.73], [623.62, 144.26]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[696.34, 291.3], [654.95, 293.96], [629.49, 285.09], [668.26, 282.83], [696.34, 164.05], [654.95, 163.65], [629.49, 164.97], [668.26, 165.31]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[730.37, 282.48], [683.44, 284.45], [659.04, 275.25], [702.31, 273.61], [730.37, 143.45], [683.44, 142.73], [659.04, 146.08], [702.31, 146.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[534.18, 238.93], [501.04, 239.84], [492.47, 236.2], [523.72, 235.4], [534.18, 141.58], [501.04, 140.97], [492.47, 143.4], [523.72, 143.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[608.08, 237.37], [573.95, 238.01], [565.38, 234.39], [597.48, 233.82], [608.08, 146.78], [573.95, 146.4], [565.38, 148.55], [597.48, 148.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[648.09, 237.46], [613.28, 238.19], [600.18, 234.21], [632.78, 233.58], [648.09, 139.51], [613.28, 138.99], [600.18, 141.85], [632.78, 142.3]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[706.08, 235.12], [676.52, 235.6], [662.12, 231.93], [689.85, 231.51], [706.08, 142.17], [676.52, 141.84], [662.12, 144.41], [689.85, 144.7]]}]\n```" - } - ] - }, - { - "id": 186, - "data_source": "Mono3DRefer", - "scene_name": "000966", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[-2.4, 384.77], [266.59, 362.7], [50.03, 910.7], [-1811.46, 1468.44], [-2.4, 123.97], [266.59, 130.08], [50.03, -21.61], [-1811.46, -175.99]]}]\n```" - } - ] - }, - { - "id": 187, - "data_source": "Mono3DRefer", - "scene_name": "000966", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[660.4, 202.05], [667.02, 201.6], [685.66, 201.89], [679.34, 202.34], [660.4, 146.15], [667.02, 146.86], [685.66, 146.4], [679.34, 145.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[304.49, 256.71], [272.51, 260.9], [241.15, 259.15], [274.07, 255.13], [304.49, 141.34], [272.51, 139.19], [241.15, 140.09], [274.07, 142.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1471.77, 665.05], [1293.29, 682.5], [1052.97, 526.84], [1183.28, 518.45], [1471.77, 118.17], [1293.29, 115.92], [1052.97, 135.95], [1183.28, 137.03]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[895.56, 329.49], [817.81, 330.17], [787.43, 311.43], [855.6, 310.91], [895.56, 139.35], [817.81, 139.16], [787.43, 144.34], [855.6, 144.49]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[812.21, 332.01], [733.78, 335.84], [701.79, 318.16], [772.19, 315.14], [812.21, 147.75], [733.78, 146.92], [701.79, 150.75], [772.19, 151.4]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1021.45, 333.71], [991.63, 342.0], [891.29, 324.08], [922.56, 317.48], [1021.45, 140.35], [991.63, 138.17], [891.29, 142.87], [922.56, 144.6]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1124.18, 337.74], [1094.57, 343.1], [979.99, 321.19], [1009.01, 317.16], [1124.18, 149.05], [1094.57, 147.98], [979.99, 152.36], [1009.01, 153.16]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[885.45, 236.85], [885.73, 235.36], [932.07, 236.64], [933.07, 238.2], [885.45, 161.23], [885.73, 161.74], [932.07, 161.3], [933.07, 160.77]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[579.45, 319.83], [523.99, 323.65], [505.31, 307.99], [555.3, 304.94], [579.45, 144.24], [523.99, 143.25], [505.31, 147.32], [555.3, 148.12]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[608.57, 300.83], [564.24, 303.25], [546.43, 291.58], [586.94, 289.59], [608.57, 158.37], [564.24, 157.92], [546.43, 160.07], [586.94, 160.44]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[645.14, 298.21], [599.45, 300.1], [582.27, 287.94], [623.62, 286.41], [645.14, 140.22], [599.45, 139.57], [582.27, 143.73], [623.62, 144.26]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[696.34, 291.3], [654.95, 293.96], [629.49, 285.09], [668.26, 282.83], [696.34, 164.05], [654.95, 163.65], [629.49, 164.97], [668.26, 165.31]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[730.37, 282.48], [683.44, 284.45], [659.04, 275.25], [702.31, 273.61], [730.37, 143.45], [683.44, 142.73], [659.04, 146.08], [702.31, 146.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[534.18, 238.93], [501.04, 239.84], [492.47, 236.2], [523.72, 235.4], [534.18, 141.58], [501.04, 140.97], [492.47, 143.4], [523.72, 143.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[608.08, 237.37], [573.95, 238.01], [565.38, 234.39], [597.48, 233.82], [608.08, 146.78], [573.95, 146.4], [565.38, 148.55], [597.48, 148.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[648.09, 237.46], [613.28, 238.19], [600.18, 234.21], [632.78, 233.58], [648.09, 139.51], [613.28, 138.99], [600.18, 141.85], [632.78, 142.3]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[706.08, 235.12], [676.52, 235.6], [662.12, 231.93], [689.85, 231.51], [706.08, 142.17], [676.52, 141.84], [662.12, 144.41], [689.85, 144.7]]}, {\"category\": \"van\", \"corners_3d\": [[-2.4, 384.77], [266.59, 362.7], [50.03, 910.7], [-1811.46, 1468.44], [-2.4, 123.97], [266.59, 130.08], [50.03, -21.61], [-1811.46, -175.99]]}]\n```" - } - ] - }, - { - "id": 188, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[1119.53, 414.15], [1006.05, 402.01], [995.81, 378.01], [1095.83, 387.59], [1119.53, 220.09], [1006.05, 218.08], [995.81, 214.1], [1095.83, 215.69]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[995.35, 444.96], [1109.54, 444.58], [1214.56, 498.14], [1077.26, 498.7], [995.35, 235.35], [1109.54, 235.27], [1214.56, 246.22], [1077.26, 246.34]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1108.11, 460.17], [1129.78, 434.39], [1287.4, 466.93], [1280.99, 500.21], [1108.11, 232.96], [1129.78, 228.2], [1287.4, 234.21], [1280.99, 240.35]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1051.72, 415.44], [973.72, 424.8], [892.92, 391.39], [963.26, 384.38], [1051.72, 239.44], [973.72, 241.76], [892.92, 233.48], [963.26, 231.75]]}]\n```" - } - ] - }, - { - "id": 189, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[464.74, 240.37], [453.43, 237.88], [492.4, 236.34], [505.07, 238.69], [464.74, 173.09], [453.43, 173.45], [492.4, 173.67], [505.07, 173.33]]}]\n```" - } - ] - }, - { - "id": 190, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[378.82, 214.43], [384.83, 217.46], [294.22, 219.08], [295.91, 215.78], [378.82, 131.92], [384.83, 127.36], [294.22, 124.92], [295.91, 129.89]]}, {\"category\": \"tram\", \"corners_3d\": [[533.38, 208.24], [552.42, 210.55], [386.75, 213.16], [380.43, 210.43], [533.38, 127.52], [552.42, 122.84], [386.75, 117.58], [380.43, 123.09]]}, {\"category\": \"tram\", \"corners_3d\": [[599.51, 201.17], [623.25, 202.8], [554.68, 203.63], [535.54, 201.87], [599.51, 123.89], [623.25, 119.1], [554.68, 116.69], [535.54, 121.84]]}, {\"category\": \"tram\", \"corners_3d\": [[820.77, 192.09], [857.27, 192.85], [625.07, 194.56], [601.25, 193.56], [820.77, 123.62], [857.27, 119.45], [625.07, 110.1], [601.25, 115.56]]}]\n```" - } - ] - }, - { - "id": 191, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[1119.53, 414.15], [1006.05, 402.01], [995.81, 378.01], [1095.83, 387.59], [1119.53, 220.09], [1006.05, 218.08], [995.81, 214.1], [1095.83, 215.69]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[464.74, 240.37], [453.43, 237.88], [492.4, 236.34], [505.07, 238.69], [464.74, 173.09], [453.43, 173.45], [492.4, 173.67], [505.07, 173.33]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[995.35, 444.96], [1109.54, 444.58], [1214.56, 498.14], [1077.26, 498.7], [995.35, 235.35], [1109.54, 235.27], [1214.56, 246.22], [1077.26, 246.34]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1108.11, 460.17], [1129.78, 434.39], [1287.4, 466.93], [1280.99, 500.21], [1108.11, 232.96], [1129.78, 228.2], [1287.4, 234.21], [1280.99, 240.35]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1051.72, 415.44], [973.72, 424.8], [892.92, 391.39], [963.26, 384.38], [1051.72, 239.44], [973.72, 241.76], [892.92, 233.48], [963.26, 231.75]]}]\n```" - } - ] - }, - { - "id": 192, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[1119.53, 414.15], [1006.05, 402.01], [995.81, 378.01], [1095.83, 387.59], [1119.53, 220.09], [1006.05, 218.08], [995.81, 214.1], [1095.83, 215.69]]}, {\"category\": \"tram\", \"corners_3d\": [[378.82, 214.43], [384.83, 217.46], [294.22, 219.08], [295.91, 215.78], [378.82, 131.92], [384.83, 127.36], [294.22, 124.92], [295.91, 129.89]]}, {\"category\": \"tram\", \"corners_3d\": [[533.38, 208.24], [552.42, 210.55], [386.75, 213.16], [380.43, 210.43], [533.38, 127.52], [552.42, 122.84], [386.75, 117.58], [380.43, 123.09]]}, {\"category\": \"tram\", \"corners_3d\": [[599.51, 201.17], [623.25, 202.8], [554.68, 203.63], [535.54, 201.87], [599.51, 123.89], [623.25, 119.1], [554.68, 116.69], [535.54, 121.84]]}, {\"category\": \"tram\", \"corners_3d\": [[820.77, 192.09], [857.27, 192.85], [625.07, 194.56], [601.25, 193.56], [820.77, 123.62], [857.27, 119.45], [625.07, 110.1], [601.25, 115.56]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[995.35, 444.96], [1109.54, 444.58], [1214.56, 498.14], [1077.26, 498.7], [995.35, 235.35], [1109.54, 235.27], [1214.56, 246.22], [1077.26, 246.34]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1108.11, 460.17], [1129.78, 434.39], [1287.4, 466.93], [1280.99, 500.21], [1108.11, 232.96], [1129.78, 228.2], [1287.4, 234.21], [1280.99, 240.35]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1051.72, 415.44], [973.72, 424.8], [892.92, 391.39], [963.26, 384.38], [1051.72, 239.44], [973.72, 241.76], [892.92, 233.48], [963.26, 231.75]]}]\n```" - } - ] - }, - { - "id": 193, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[464.74, 240.37], [453.43, 237.88], [492.4, 236.34], [505.07, 238.69], [464.74, 173.09], [453.43, 173.45], [492.4, 173.67], [505.07, 173.33]]}, {\"category\": \"tram\", \"corners_3d\": [[378.82, 214.43], [384.83, 217.46], [294.22, 219.08], [295.91, 215.78], [378.82, 131.92], [384.83, 127.36], [294.22, 124.92], [295.91, 129.89]]}, {\"category\": \"tram\", \"corners_3d\": [[533.38, 208.24], [552.42, 210.55], [386.75, 213.16], [380.43, 210.43], [533.38, 127.52], [552.42, 122.84], [386.75, 117.58], [380.43, 123.09]]}, {\"category\": \"tram\", \"corners_3d\": [[599.51, 201.17], [623.25, 202.8], [554.68, 203.63], [535.54, 201.87], [599.51, 123.89], [623.25, 119.1], [554.68, 116.69], [535.54, 121.84]]}, {\"category\": \"tram\", \"corners_3d\": [[820.77, 192.09], [857.27, 192.85], [625.07, 194.56], [601.25, 193.56], [820.77, 123.62], [857.27, 119.45], [625.07, 110.1], [601.25, 115.56]]}]\n```" - } - ] - }, - { - "id": 194, - "data_source": "Mono3DRefer", - "scene_name": "000971", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting, pedestrian, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[1119.53, 414.15], [1006.05, 402.01], [995.81, 378.01], [1095.83, 387.59], [1119.53, 220.09], [1006.05, 218.08], [995.81, 214.1], [1095.83, 215.69]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[464.74, 240.37], [453.43, 237.88], [492.4, 236.34], [505.07, 238.69], [464.74, 173.09], [453.43, 173.45], [492.4, 173.67], [505.07, 173.33]]}, {\"category\": \"tram\", \"corners_3d\": [[378.82, 214.43], [384.83, 217.46], [294.22, 219.08], [295.91, 215.78], [378.82, 131.92], [384.83, 127.36], [294.22, 124.92], [295.91, 129.89]]}, {\"category\": \"tram\", \"corners_3d\": [[533.38, 208.24], [552.42, 210.55], [386.75, 213.16], [380.43, 210.43], [533.38, 127.52], [552.42, 122.84], [386.75, 117.58], [380.43, 123.09]]}, {\"category\": \"tram\", \"corners_3d\": [[599.51, 201.17], [623.25, 202.8], [554.68, 203.63], [535.54, 201.87], [599.51, 123.89], [623.25, 119.1], [554.68, 116.69], [535.54, 121.84]]}, {\"category\": \"tram\", \"corners_3d\": [[820.77, 192.09], [857.27, 192.85], [625.07, 194.56], [601.25, 193.56], [820.77, 123.62], [857.27, 119.45], [625.07, 110.1], [601.25, 115.56]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[995.35, 444.96], [1109.54, 444.58], [1214.56, 498.14], [1077.26, 498.7], [995.35, 235.35], [1109.54, 235.27], [1214.56, 246.22], [1077.26, 246.34]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1108.11, 460.17], [1129.78, 434.39], [1287.4, 466.93], [1280.99, 500.21], [1108.11, 232.96], [1129.78, 228.2], [1287.4, 234.21], [1280.99, 240.35]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[1051.72, 415.44], [973.72, 424.8], [892.92, 391.39], [963.26, 384.38], [1051.72, 239.44], [973.72, 241.76], [892.92, 233.48], [963.26, 231.75]]}]\n```" - } - ] - }, - { - "id": 195, - "data_source": "Mono3DRefer", - "scene_name": "000988", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[242.61, 226.24], [198.9, 226.24], [241.1, 220.75], [280.31, 220.75], [242.61, 192.75], [198.9, 192.75], [241.1, 190.7], [280.31, 190.7]]}, {\"category\": \"car\", \"corners_3d\": [[410.27, 203.34], [388.55, 203.16], [409.17, 201.38], [429.72, 201.54], [410.27, 183.86], [388.55, 183.8], [409.17, 183.16], [429.72, 183.21]]}, {\"category\": \"car\", \"corners_3d\": [[-54.26, 261.77], [-131.01, 261.68], [15.75, 244.23], [77.53, 244.29], [-54.26, 196.17], [-131.01, 196.15], [15.75, 191.57], [77.53, 191.59]]}]\n```" - } - ] - }, - { - "id": 196, - "data_source": "Mono3DRefer", - "scene_name": "000988", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[363.89, 203.16], [356.12, 203.12], [366.11, 202.25], [373.67, 202.29], [363.89, 179.73], [356.12, 179.72], [366.11, 179.53], [373.67, 179.53]]}]\n```" - } - ] - }, - { - "id": 197, - "data_source": "Mono3DRefer", - "scene_name": "000988", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[242.61, 226.24], [198.9, 226.24], [241.1, 220.75], [280.31, 220.75], [242.61, 192.75], [198.9, 192.75], [241.1, 190.7], [280.31, 190.7]]}, {\"category\": \"cyclist\", \"corners_3d\": [[363.89, 203.16], [356.12, 203.12], [366.11, 202.25], [373.67, 202.29], [363.89, 179.73], [356.12, 179.72], [366.11, 179.53], [373.67, 179.53]]}, {\"category\": \"car\", \"corners_3d\": [[410.27, 203.34], [388.55, 203.16], [409.17, 201.38], [429.72, 201.54], [410.27, 183.86], [388.55, 183.8], [409.17, 183.16], [429.72, 183.21]]}, {\"category\": \"car\", \"corners_3d\": [[-54.26, 261.77], [-131.01, 261.68], [15.75, 244.23], [77.53, 244.29], [-54.26, 196.17], [-131.01, 196.15], [15.75, 191.57], [77.53, 191.59]]}]\n```" - } - ] - }, - { - "id": 198, - "data_source": "Mono3DRefer", - "scene_name": "000991", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[611.2, 212.92], [647.59, 212.99], [649.06, 216.7], [609.3, 216.62], [611.2, 176.34], [647.59, 176.35], [649.06, 176.67], [609.3, 176.66]]}, {\"category\": \"car\", \"corners_3d\": [[485.65, 249.08], [416.66, 248.72], [460.95, 234.01], [516.68, 234.25], [485.65, 189.5], [416.66, 189.42], [460.95, 186.21], [516.68, 186.26]]}, {\"category\": \"car\", \"corners_3d\": [[528.87, 229.03], [481.12, 229.14], [497.1, 220.67], [537.66, 220.59], [528.87, 185.4], [481.12, 185.42], [497.1, 183.53], [537.66, 183.51]]}, {\"category\": \"car\", \"corners_3d\": [[551.4, 208.02], [519.15, 207.98], [530.41, 204.43], [559.41, 204.47], [551.4, 180.44], [519.15, 180.43], [530.41, 179.67], [559.41, 179.68]]}, {\"category\": \"car\", \"corners_3d\": [[567.23, 195.27], [545.94, 195.23], [553.37, 193.66], [573.18, 193.7], [567.23, 174.39], [545.94, 174.39], [553.37, 174.28], [573.18, 174.28]]}]\n```" - } - ] - }, - { - "id": 199, - "data_source": "Mono3DRefer", - "scene_name": "001019", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[513.43, 200.07], [501.86, 200.06], [503.17, 199.77], [514.62, 199.77], [513.43, 177.45], [501.86, 177.44], [503.17, 177.4], [514.62, 177.4]]}]\n```" - } - ] - }, - { - "id": 200, - "data_source": "Mono3DRefer", - "scene_name": "001065", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[584.63, 184.83], [578.69, 184.83], [578.89, 184.55], [584.7, 184.55], [584.63, 168.55], [578.69, 168.55], [578.89, 168.65], [584.7, 168.65]]}]\n```" - } - ] - }, - { - "id": 201, - "data_source": "Mono3DRefer", - "scene_name": "001065", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[747.31, 278.65], [854.1, 278.19], [1011.19, 340.48], [841.89, 341.65], [747.31, 183.97], [854.1, 183.92], [1011.19, 190.46], [841.89, 190.59]]}, {\"category\": \"car\", \"corners_3d\": [[682.59, 231.8], [737.74, 231.63], [770.76, 244.03], [704.05, 244.29], [682.59, 170.96], [737.74, 170.97], [770.76, 170.57], [704.05, 170.56]]}, {\"category\": \"car\", \"corners_3d\": [[437.46, 241.89], [413.45, 249.58], [226.23, 248.36], [268.39, 240.9], [437.46, 175.82], [413.45, 176.15], [226.23, 176.1], [268.39, 175.78]]}, {\"category\": \"car\", \"corners_3d\": [[312.3, 229.14], [334.18, 224.88], [444.1, 224.78], [431.25, 229.03], [312.3, 160.09], [334.18, 161.06], [444.1, 161.08], [431.25, 160.12]]}, {\"category\": \"car\", \"corners_3d\": [[500.19, 212.97], [493.46, 215.29], [394.36, 215.23], [406.47, 212.92], [500.19, 175.22], [493.46, 175.36], [394.36, 175.35], [406.47, 175.22]]}, {\"category\": \"car\", \"corners_3d\": [[517.56, 209.82], [511.91, 211.92], [419.09, 211.87], [429.74, 209.77], [517.56, 172.6], [511.91, 172.59], [419.09, 172.59], [429.74, 172.6]]}, {\"category\": \"car\", \"corners_3d\": [[541.04, 192.64], [538.99, 193.39], [472.25, 193.43], [476.75, 192.68], [541.04, 162.15], [538.99, 161.74], [472.25, 161.72], [476.75, 162.13]]}, {\"category\": \"car\", \"corners_3d\": [[651.32, 205.49], [683.85, 205.45], [694.74, 209.17], [658.5, 209.23], [651.32, 175.26], [683.85, 175.25], [694.74, 175.53], [658.5, 175.53]]}, {\"category\": \"car\", \"corners_3d\": [[625.8, 186.08], [642.52, 186.06], [646.94, 186.91], [629.15, 186.93], [625.8, 170.91], [642.52, 170.92], [646.94, 170.79], [629.15, 170.79]]}, {\"category\": \"car\", \"corners_3d\": [[933.17, 442.84], [1223.42, 442.93], [4529.63, 1898.91], [2671.09, 1895.39], [933.17, 166.13], [1223.42, 166.12], [4529.63, 129.84], [2671.09, 129.93]]}]\n```" - } - ] - }, - { - "id": 202, - "data_source": "Mono3DRefer", - "scene_name": "001065", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[747.31, 278.65], [854.1, 278.19], [1011.19, 340.48], [841.89, 341.65], [747.31, 183.97], [854.1, 183.92], [1011.19, 190.46], [841.89, 190.59]]}, {\"category\": \"car\", \"corners_3d\": [[682.59, 231.8], [737.74, 231.63], [770.76, 244.03], [704.05, 244.29], [682.59, 170.96], [737.74, 170.97], [770.76, 170.57], [704.05, 170.56]]}, {\"category\": \"car\", \"corners_3d\": [[437.46, 241.89], [413.45, 249.58], [226.23, 248.36], [268.39, 240.9], [437.46, 175.82], [413.45, 176.15], [226.23, 176.1], [268.39, 175.78]]}, {\"category\": \"car\", \"corners_3d\": [[312.3, 229.14], [334.18, 224.88], [444.1, 224.78], [431.25, 229.03], [312.3, 160.09], [334.18, 161.06], [444.1, 161.08], [431.25, 160.12]]}, {\"category\": \"car\", \"corners_3d\": [[500.19, 212.97], [493.46, 215.29], [394.36, 215.23], [406.47, 212.92], [500.19, 175.22], [493.46, 175.36], [394.36, 175.35], [406.47, 175.22]]}, {\"category\": \"car\", \"corners_3d\": [[517.56, 209.82], [511.91, 211.92], [419.09, 211.87], [429.74, 209.77], [517.56, 172.6], [511.91, 172.59], [419.09, 172.59], [429.74, 172.6]]}, {\"category\": \"car\", \"corners_3d\": [[541.04, 192.64], [538.99, 193.39], [472.25, 193.43], [476.75, 192.68], [541.04, 162.15], [538.99, 161.74], [472.25, 161.72], [476.75, 162.13]]}, {\"category\": \"cyclist\", \"corners_3d\": [[584.63, 184.83], [578.69, 184.83], [578.89, 184.55], [584.7, 184.55], [584.63, 168.55], [578.69, 168.55], [578.89, 168.65], [584.7, 168.65]]}, {\"category\": \"car\", \"corners_3d\": [[651.32, 205.49], [683.85, 205.45], [694.74, 209.17], [658.5, 209.23], [651.32, 175.26], [683.85, 175.25], [694.74, 175.53], [658.5, 175.53]]}, {\"category\": \"car\", \"corners_3d\": [[625.8, 186.08], [642.52, 186.06], [646.94, 186.91], [629.15, 186.93], [625.8, 170.91], [642.52, 170.92], [646.94, 170.79], [629.15, 170.79]]}, {\"category\": \"car\", \"corners_3d\": [[933.17, 442.84], [1223.42, 442.93], [4529.63, 1898.91], [2671.09, 1895.39], [933.17, 166.13], [1223.42, 166.12], [4529.63, 129.84], [2671.09, 129.93]]}]\n```" - } - ] - }, - { - "id": 203, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[306.48, 412.34], [212.95, 414.97], [302.42, 350.11], [369.45, 348.73], [306.48, 153.86], [212.95, 153.55], [302.42, 161.32], [369.45, 161.48]]}]\n```" - } - ] - }, - { - "id": 204, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[648.23, 330.75], [730.13, 331.27], [744.32, 351.02], [651.61, 350.35], [648.23, 162.94], [730.13, 162.87], [744.32, 160.41], [651.61, 160.5]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[702.36, 328.41], [787.97, 328.6], [813.8, 349.72], [715.87, 349.48], [702.36, 156.71], [787.97, 156.67], [813.8, 153.11], [715.87, 153.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[213.08, 324.87], [300.71, 324.2], [259.72, 345.73], [158.58, 346.62], [213.08, 157.93], [300.71, 158.04], [259.72, 154.49], [158.58, 154.35]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[765.37, 232.13], [798.19, 231.87], [814.47, 234.84], [779.79, 235.13], [765.37, 161.26], [798.19, 161.36], [814.47, 160.18], [779.79, 160.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[820.62, 212.89], [802.16, 212.19], [811.93, 211.73], [830.33, 212.41], [820.62, 159.06], [802.16, 159.57], [811.93, 159.9], [830.33, 159.41]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[832.69, 249.18], [867.37, 250.13], [869.63, 254.42], [832.78, 253.34], [832.69, 162.23], [867.37, 161.96], [869.63, 160.74], [832.78, 161.04]]}]\n```" - } - ] - }, - { - "id": 205, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[557.1, 208.79], [516.48, 208.78], [526.76, 205.68], [562.77, 205.69], [557.1, 166.73], [516.48, 166.74], [526.76, 168.42], [562.77, 168.41]]}]\n```" - } - ] - }, - { - "id": 206, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[648.23, 330.75], [730.13, 331.27], [744.32, 351.02], [651.61, 350.35], [648.23, 162.94], [730.13, 162.87], [744.32, 160.41], [651.61, 160.5]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[702.36, 328.41], [787.97, 328.6], [813.8, 349.72], [715.87, 349.48], [702.36, 156.71], [787.97, 156.67], [813.8, 153.11], [715.87, 153.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[213.08, 324.87], [300.71, 324.2], [259.72, 345.73], [158.58, 346.62], [213.08, 157.93], [300.71, 158.04], [259.72, 154.49], [158.58, 154.35]]}, {\"category\": \"cyclist\", \"corners_3d\": [[306.48, 412.34], [212.95, 414.97], [302.42, 350.11], [369.45, 348.73], [306.48, 153.86], [212.95, 153.55], [302.42, 161.32], [369.45, 161.48]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[765.37, 232.13], [798.19, 231.87], [814.47, 234.84], [779.79, 235.13], [765.37, 161.26], [798.19, 161.36], [814.47, 160.18], [779.79, 160.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[820.62, 212.89], [802.16, 212.19], [811.93, 211.73], [830.33, 212.41], [820.62, 159.06], [802.16, 159.57], [811.93, 159.9], [830.33, 159.41]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[832.69, 249.18], [867.37, 250.13], [869.63, 254.42], [832.78, 253.34], [832.69, 162.23], [867.37, 161.96], [869.63, 160.74], [832.78, 161.04]]}]\n```" - } - ] - }, - { - "id": 207, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[306.48, 412.34], [212.95, 414.97], [302.42, 350.11], [369.45, 348.73], [306.48, 153.86], [212.95, 153.55], [302.42, 161.32], [369.45, 161.48]]}, {\"category\": \"car\", \"corners_3d\": [[557.1, 208.79], [516.48, 208.78], [526.76, 205.68], [562.77, 205.69], [557.1, 166.73], [516.48, 166.74], [526.76, 168.42], [562.77, 168.41]]}]\n```" - } - ] - }, - { - "id": 208, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[648.23, 330.75], [730.13, 331.27], [744.32, 351.02], [651.61, 350.35], [648.23, 162.94], [730.13, 162.87], [744.32, 160.41], [651.61, 160.5]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[702.36, 328.41], [787.97, 328.6], [813.8, 349.72], [715.87, 349.48], [702.36, 156.71], [787.97, 156.67], [813.8, 153.11], [715.87, 153.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[213.08, 324.87], [300.71, 324.2], [259.72, 345.73], [158.58, 346.62], [213.08, 157.93], [300.71, 158.04], [259.72, 154.49], [158.58, 154.35]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[765.37, 232.13], [798.19, 231.87], [814.47, 234.84], [779.79, 235.13], [765.37, 161.26], [798.19, 161.36], [814.47, 160.18], [779.79, 160.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[820.62, 212.89], [802.16, 212.19], [811.93, 211.73], [830.33, 212.41], [820.62, 159.06], [802.16, 159.57], [811.93, 159.9], [830.33, 159.41]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[832.69, 249.18], [867.37, 250.13], [869.63, 254.42], [832.78, 253.34], [832.69, 162.23], [867.37, 161.96], [869.63, 160.74], [832.78, 161.04]]}, {\"category\": \"car\", \"corners_3d\": [[557.1, 208.79], [516.48, 208.78], [526.76, 205.68], [562.77, 205.69], [557.1, 166.73], [516.48, 166.74], [526.76, 168.42], [562.77, 168.41]]}]\n```" - } - ] - }, - { - "id": 209, - "data_source": "Mono3DRefer", - "scene_name": "001068", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[648.23, 330.75], [730.13, 331.27], [744.32, 351.02], [651.61, 350.35], [648.23, 162.94], [730.13, 162.87], [744.32, 160.41], [651.61, 160.5]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[702.36, 328.41], [787.97, 328.6], [813.8, 349.72], [715.87, 349.48], [702.36, 156.71], [787.97, 156.67], [813.8, 153.11], [715.87, 153.15]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[213.08, 324.87], [300.71, 324.2], [259.72, 345.73], [158.58, 346.62], [213.08, 157.93], [300.71, 158.04], [259.72, 154.49], [158.58, 154.35]]}, {\"category\": \"cyclist\", \"corners_3d\": [[306.48, 412.34], [212.95, 414.97], [302.42, 350.11], [369.45, 348.73], [306.48, 153.86], [212.95, 153.55], [302.42, 161.32], [369.45, 161.48]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[765.37, 232.13], [798.19, 231.87], [814.47, 234.84], [779.79, 235.13], [765.37, 161.26], [798.19, 161.36], [814.47, 160.18], [779.79, 160.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[820.62, 212.89], [802.16, 212.19], [811.93, 211.73], [830.33, 212.41], [820.62, 159.06], [802.16, 159.57], [811.93, 159.9], [830.33, 159.41]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[832.69, 249.18], [867.37, 250.13], [869.63, 254.42], [832.78, 253.34], [832.69, 162.23], [867.37, 161.96], [869.63, 160.74], [832.78, 161.04]]}, {\"category\": \"car\", \"corners_3d\": [[557.1, 208.79], [516.48, 208.78], [526.76, 205.68], [562.77, 205.69], [557.1, 166.73], [516.48, 166.74], [526.76, 168.42], [562.77, 168.41]]}]\n```" - } - ] - }, - { - "id": 210, - "data_source": "Mono3DRefer", - "scene_name": "001084", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[499.67, 268.5], [385.69, 268.36], [445.33, 243.65], [529.85, 243.73], [499.67, 153.57], [385.69, 153.6], [445.33, 158.58], [529.85, 158.57]]}, {\"category\": \"van\", \"corners_3d\": [[709.84, 252.72], [787.83, 252.73], [846.84, 279.26], [742.95, 279.25], [709.84, 168.05], [787.83, 168.05], [846.84, 166.46], [742.95, 166.46]]}, {\"category\": \"van\", \"corners_3d\": [[572.35, 215.27], [525.32, 215.22], [539.35, 209.16], [579.66, 209.19], [572.35, 163.97], [525.32, 163.98], [539.35, 165.25], [579.66, 165.24]]}]\n```" - } - ] - }, - { - "id": 211, - "data_source": "Mono3DRefer", - "scene_name": "001084", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[559.97, 230.2], [501.53, 230.2], [521.06, 219.78], [568.88, 219.78], [559.97, 177.86], [501.53, 177.86], [521.06, 176.95], [568.88, 176.95]]}, {\"category\": \"car\", \"corners_3d\": [[585.57, 205.79], [544.38, 205.57], [563.01, 201.55], [599.22, 201.72], [585.57, 166.62], [544.38, 166.67], [563.01, 167.43], [599.22, 167.39]]}, {\"category\": \"car\", \"corners_3d\": [[639.98, 193.23], [652.89, 193.85], [616.22, 194.45], [604.06, 193.79], [639.98, 168.8], [652.89, 168.68], [616.22, 168.56], [604.06, 168.69]]}, {\"category\": \"car\", \"corners_3d\": [[-81.4, 301.94], [26.43, 280.65], [334.16, 278.37], [288.3, 298.69], [-81.4, 177.07], [26.43, 176.38], [334.16, 176.3], [288.3, 176.96]]}]\n```" - } - ] - }, - { - "id": 212, - "data_source": "Mono3DRefer", - "scene_name": "001084", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[499.67, 268.5], [385.69, 268.36], [445.33, 243.65], [529.85, 243.73], [499.67, 153.57], [385.69, 153.6], [445.33, 158.58], [529.85, 158.57]]}, {\"category\": \"van\", \"corners_3d\": [[709.84, 252.72], [787.83, 252.73], [846.84, 279.26], [742.95, 279.25], [709.84, 168.05], [787.83, 168.05], [846.84, 166.46], [742.95, 166.46]]}, {\"category\": \"car\", \"corners_3d\": [[559.97, 230.2], [501.53, 230.2], [521.06, 219.78], [568.88, 219.78], [559.97, 177.86], [501.53, 177.86], [521.06, 176.95], [568.88, 176.95]]}, {\"category\": \"van\", \"corners_3d\": [[572.35, 215.27], [525.32, 215.22], [539.35, 209.16], [579.66, 209.19], [572.35, 163.97], [525.32, 163.98], [539.35, 165.25], [579.66, 165.24]]}, {\"category\": \"car\", \"corners_3d\": [[585.57, 205.79], [544.38, 205.57], [563.01, 201.55], [599.22, 201.72], [585.57, 166.62], [544.38, 166.67], [563.01, 167.43], [599.22, 167.39]]}, {\"category\": \"car\", \"corners_3d\": [[639.98, 193.23], [652.89, 193.85], [616.22, 194.45], [604.06, 193.79], [639.98, 168.8], [652.89, 168.68], [616.22, 168.56], [604.06, 168.69]]}, {\"category\": \"car\", \"corners_3d\": [[-81.4, 301.94], [26.43, 280.65], [334.16, 278.37], [288.3, 298.69], [-81.4, 177.07], [26.43, 176.38], [334.16, 176.3], [288.3, 176.96]]}]\n```" - } - ] - }, - { - "id": 213, - "data_source": "Mono3DRefer", - "scene_name": "001089", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[366.09, 216.48], [318.23, 217.19], [331.86, 211.07], [373.02, 210.54], [366.09, 160.04], [318.23, 159.83], [331.86, 161.63], [373.02, 161.78]]}]\n```" - } - ] - }, - { - "id": 214, - "data_source": "Mono3DRefer", - "scene_name": "001089", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[247.07, 366.05], [-16.67, 386.73], [152.67, 291.03], [295.53, 284.43], [247.07, 137.58], [-16.67, 133.8], [152.67, 151.28], [295.53, 152.48]]}, {\"category\": \"car\", \"corners_3d\": [[311.0, 227.76], [356.82, 226.78], [353.56, 234.69], [300.8, 235.99], [311.0, 185.02], [356.82, 184.8], [353.56, 186.56], [300.8, 186.84]]}, {\"category\": \"car\", \"corners_3d\": [[493.91, 224.73], [535.82, 223.91], [552.95, 230.48], [505.86, 231.53], [493.91, 182.1], [535.82, 181.95], [552.95, 183.12], [505.86, 183.31]]}, {\"category\": \"car\", \"corners_3d\": [[471.01, 219.71], [508.35, 218.96], [525.1, 224.58], [483.42, 225.53], [471.01, 183.31], [508.35, 183.14], [525.1, 184.4], [483.42, 184.61]]}, {\"category\": \"car\", \"corners_3d\": [[455.82, 206.64], [482.84, 206.38], [485.91, 209.41], [456.46, 209.73], [455.82, 182.34], [482.84, 182.26], [485.91, 183.11], [456.46, 183.2]]}, {\"category\": \"car\", \"corners_3d\": [[516.13, 238.45], [577.23, 237.08], [607.74, 250.03], [534.78, 252.01], [516.13, 181.54], [577.23, 181.36], [607.74, 183.07], [534.78, 183.34]]}]\n```" - } - ] - }, - { - "id": 215, - "data_source": "Mono3DRefer", - "scene_name": "001089", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[247.07, 366.05], [-16.67, 386.73], [152.67, 291.03], [295.53, 284.43], [247.07, 137.58], [-16.67, 133.8], [152.67, 151.28], [295.53, 152.48]]}, {\"category\": \"car\", \"corners_3d\": [[311.0, 227.76], [356.82, 226.78], [353.56, 234.69], [300.8, 235.99], [311.0, 185.02], [356.82, 184.8], [353.56, 186.56], [300.8, 186.84]]}, {\"category\": \"car\", \"corners_3d\": [[493.91, 224.73], [535.82, 223.91], [552.95, 230.48], [505.86, 231.53], [493.91, 182.1], [535.82, 181.95], [552.95, 183.12], [505.86, 183.31]]}, {\"category\": \"car\", \"corners_3d\": [[471.01, 219.71], [508.35, 218.96], [525.1, 224.58], [483.42, 225.53], [471.01, 183.31], [508.35, 183.14], [525.1, 184.4], [483.42, 184.61]]}, {\"category\": \"truck\", \"corners_3d\": [[366.09, 216.48], [318.23, 217.19], [331.86, 211.07], [373.02, 210.54], [366.09, 160.04], [318.23, 159.83], [331.86, 161.63], [373.02, 161.78]]}, {\"category\": \"car\", \"corners_3d\": [[455.82, 206.64], [482.84, 206.38], [485.91, 209.41], [456.46, 209.73], [455.82, 182.34], [482.84, 182.26], [485.91, 183.11], [456.46, 183.2]]}, {\"category\": \"car\", \"corners_3d\": [[516.13, 238.45], [577.23, 237.08], [607.74, 250.03], [534.78, 252.01], [516.13, 181.54], [577.23, 181.36], [607.74, 183.07], [534.78, 183.34]]}]\n```" - } - ] - }, - { - "id": 216, - "data_source": "Mono3DRefer", - "scene_name": "001097", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-282.84, 743.96], [-812.94, 735.7], [91.01, 381.48], [290.57, 382.61], [-282.84, 223.14], [-812.94, 222.41], [91.01, 191.22], [290.57, 191.32]]}, {\"category\": \"car\", \"corners_3d\": [[516.61, 236.79], [592.38, 237.34], [564.96, 260.73], [462.35, 259.72], [516.61, 168.21], [592.38, 168.17], [564.96, 166.47], [462.35, 166.55]]}, {\"category\": \"car\", \"corners_3d\": [[838.53, 257.34], [936.1, 261.09], [955.58, 288.59], [828.2, 282.23], [838.53, 172.25], [936.1, 172.23], [955.58, 172.03], [828.2, 172.08]]}, {\"category\": \"car\", \"corners_3d\": [[401.97, 260.74], [315.14, 260.75], [384.79, 239.91], [451.04, 239.91], [401.97, 182.72], [315.14, 182.72], [384.79, 180.38], [451.04, 180.38]]}, {\"category\": \"car\", \"corners_3d\": [[952.97, 227.3], [1020.15, 231.24], [929.18, 242.27], [857.53, 236.77], [952.97, 165.41], [1020.15, 164.87], [929.18, 163.37], [857.53, 164.12]]}]\n```" - } - ] - }, - { - "id": 217, - "data_source": "Mono3DRefer", - "scene_name": "001115", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[592.43, 188.56], [615.51, 188.55], [617.73, 191.32], [590.58, 191.33], [592.43, 163.55], [615.51, 163.55], [617.73, 161.91], [590.58, 161.91]]}]\n```" - } - ] - }, - { - "id": 218, - "data_source": "Mono3DRefer", - "scene_name": "001115", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[293.28, 215.72], [260.21, 215.73], [280.31, 213.25], [311.47, 213.25], [293.28, 188.21], [260.21, 188.21], [280.31, 187.33], [311.47, 187.33]]}]\n```" - } - ] - }, - { - "id": 219, - "data_source": "Mono3DRefer", - "scene_name": "001115", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[592.43, 188.56], [615.51, 188.55], [617.73, 191.32], [590.58, 191.33], [592.43, 163.55], [615.51, 163.55], [617.73, 161.91], [590.58, 161.91]]}, {\"category\": \"car\", \"corners_3d\": [[293.28, 215.72], [260.21, 215.73], [280.31, 213.25], [311.47, 213.25], [293.28, 188.21], [260.21, 188.21], [280.31, 187.33], [311.47, 187.33]]}]\n```" - } - ] - }, - { - "id": 220, - "data_source": "Mono3DRefer", - "scene_name": "001139", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[1228.6, 184.57], [1182.68, 184.62], [1098.82, 183.02], [1138.83, 182.99], [1228.6, 129.52], [1182.68, 129.34], [1098.82, 135.26], [1138.83, 135.39]]}]\n```" - } - ] - }, - { - "id": 221, - "data_source": "Mono3DRefer", - "scene_name": "001139", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[488.83, 221.29], [450.39, 221.29], [468.97, 215.62], [502.9, 215.62], [488.83, 180.76], [450.39, 180.77], [468.97, 179.84], [502.9, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[505.57, 213.96], [474.26, 213.94], [487.86, 210.0], [516.17, 210.02], [505.57, 183.17], [474.26, 183.17], [487.86, 182.18], [516.17, 182.18]]}, {\"category\": \"car\", \"corners_3d\": [[65.68, 240.62], [108.19, 240.67], [12.95, 253.35], [-37.44, 253.29], [65.68, 203.08], [108.19, 203.1], [12.95, 208.76], [-37.44, 208.73]]}, {\"category\": \"car\", \"corners_3d\": [[387.96, 257.06], [323.79, 257.14], [370.4, 243.01], [423.77, 242.95], [387.96, 198.15], [323.79, 198.17], [370.4, 193.93], [423.77, 193.91]]}, {\"category\": \"car\", \"corners_3d\": [[670.06, 219.69], [721.3, 219.69], [741.45, 228.18], [680.92, 228.18], [670.06, 175.48], [721.3, 175.48], [741.45, 175.95], [680.92, 175.95]]}, {\"category\": \"car\", \"corners_3d\": [[439.07, 237.61], [391.04, 237.57], [425.0, 227.81], [465.81, 227.84], [439.07, 194.22], [391.04, 194.21], [425.0, 190.98], [465.81, 190.99]]}]\n```" - } - ] - }, - { - "id": 222, - "data_source": "Mono3DRefer", - "scene_name": "001139", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[488.83, 221.29], [450.39, 221.29], [468.97, 215.62], [502.9, 215.62], [488.83, 180.76], [450.39, 180.77], [468.97, 179.84], [502.9, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[505.57, 213.96], [474.26, 213.94], [487.86, 210.0], [516.17, 210.02], [505.57, 183.17], [474.26, 183.17], [487.86, 182.18], [516.17, 182.18]]}, {\"category\": \"van\", \"corners_3d\": [[1228.6, 184.57], [1182.68, 184.62], [1098.82, 183.02], [1138.83, 182.99], [1228.6, 129.52], [1182.68, 129.34], [1098.82, 135.26], [1138.83, 135.39]]}, {\"category\": \"car\", \"corners_3d\": [[65.68, 240.62], [108.19, 240.67], [12.95, 253.35], [-37.44, 253.29], [65.68, 203.08], [108.19, 203.1], [12.95, 208.76], [-37.44, 208.73]]}, {\"category\": \"car\", \"corners_3d\": [[387.96, 257.06], [323.79, 257.14], [370.4, 243.01], [423.77, 242.95], [387.96, 198.15], [323.79, 198.17], [370.4, 193.93], [423.77, 193.91]]}, {\"category\": \"car\", \"corners_3d\": [[670.06, 219.69], [721.3, 219.69], [741.45, 228.18], [680.92, 228.18], [670.06, 175.48], [721.3, 175.48], [741.45, 175.95], [680.92, 175.95]]}, {\"category\": \"car\", \"corners_3d\": [[439.07, 237.61], [391.04, 237.57], [425.0, 227.81], [465.81, 227.84], [439.07, 194.22], [391.04, 194.21], [425.0, 190.98], [465.81, 190.99]]}]\n```" - } - ] - }, - { - "id": 223, - "data_source": "Mono3DRefer", - "scene_name": "001140", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[186.46, 327.3], [60.89, 327.59], [213.49, 283.93], [303.48, 283.79], [186.46, 212.46], [60.89, 212.53], [213.49, 201.34], [303.48, 201.3]]}, {\"category\": \"car\", \"corners_3d\": [[707.83, 242.95], [786.5, 242.96], [840.82, 264.55], [737.91, 264.54], [707.83, 175.09], [786.5, 175.09], [840.82, 175.77], [737.91, 175.77]]}, {\"category\": \"car\", \"corners_3d\": [[348.98, 267.52], [273.26, 267.43], [348.35, 246.72], [407.53, 246.78], [348.98, 199.03], [273.26, 199.0], [348.35, 193.28], [407.53, 193.29]]}, {\"category\": \"car\", \"corners_3d\": [[448.42, 235.91], [395.58, 235.91], [428.5, 226.19], [473.18, 226.18], [448.42, 180.21], [395.58, 180.21], [428.5, 179.07], [473.18, 179.07]]}, {\"category\": \"car\", \"corners_3d\": [[480.55, 223.28], [440.2, 223.22], [462.25, 217.16], [497.77, 217.2], [480.55, 183.48], [440.2, 183.46], [462.25, 182.19], [497.77, 182.2]]}, {\"category\": \"car\", \"corners_3d\": [[638.19, 193.09], [656.46, 193.1], [659.1, 194.47], [639.6, 194.46], [638.19, 176.04], [656.46, 176.04], [659.1, 176.26], [639.6, 176.26]]}]\n```" - } - ] - }, - { - "id": 224, - "data_source": "Mono3DRefer", - "scene_name": "001144", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[567.17, 240.5], [528.22, 240.58], [536.34, 232.29], [570.52, 232.23], [567.17, 204.51], [528.22, 204.55], [536.34, 200.67], [570.52, 200.64]]}]\n```" - } - ] - }, - { - "id": 225, - "data_source": "Mono3DRefer", - "scene_name": "001177", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[808.49, 217.54], [825.55, 217.58], [840.16, 221.08], [821.74, 221.03], [808.49, 163.08], [825.55, 163.07], [840.16, 162.3], [821.74, 162.31]]}]\n```" - } - ] - }, - { - "id": 226, - "data_source": "Mono3DRefer", - "scene_name": "001177", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[389.73, 210.28], [358.07, 210.18], [383.34, 206.97], [412.33, 207.06], [389.73, 183.99], [358.07, 183.96], [383.34, 183.01], [412.33, 183.03]]}]\n```" - } - ] - }, - { - "id": 227, - "data_source": "Mono3DRefer", - "scene_name": "001177", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[808.49, 217.54], [825.55, 217.58], [840.16, 221.08], [821.74, 221.03], [808.49, 163.08], [825.55, 163.07], [840.16, 162.3], [821.74, 162.31]]}, {\"category\": \"car\", \"corners_3d\": [[389.73, 210.28], [358.07, 210.18], [383.34, 206.97], [412.33, 207.06], [389.73, 183.99], [358.07, 183.96], [383.34, 183.01], [412.33, 183.03]]}]\n```" - } - ] - }, - { - "id": 228, - "data_source": "Mono3DRefer", - "scene_name": "001228", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[1050.73, 298.12], [1082.5, 306.1], [988.05, 306.62], [962.18, 298.57], [1050.73, 157.47], [1082.5, 155.9], [988.05, 155.8], [962.18, 157.38]]}]\n```" - } - ] - }, - { - "id": 229, - "data_source": "Mono3DRefer", - "scene_name": "001236", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[247.83, 223.73], [204.97, 223.73], [253.12, 217.66], [290.87, 217.66], [247.83, 187.45], [204.97, 187.45], [253.12, 185.71], [290.87, 185.71]]}, {\"category\": \"car\", \"corners_3d\": [[357.34, 208.32], [327.32, 208.29], [349.09, 205.68], [376.92, 205.71], [357.34, 185.12], [327.32, 185.11], [349.09, 184.21], [376.92, 184.22]]}]\n```" - } - ] - }, - { - "id": 230, - "data_source": "Mono3DRefer", - "scene_name": "001243", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[598.79, 219.25], [639.82, 219.33], [640.96, 225.91], [594.13, 225.8], [598.79, 178.74], [639.82, 178.75], [640.96, 179.58], [594.13, 179.57]]}, {\"category\": \"car\", \"corners_3d\": [[258.94, 259.83], [175.93, 259.63], [281.2, 239.22], [344.82, 239.33], [258.94, 189.83], [175.93, 189.8], [281.2, 185.81], [344.82, 185.83]]}, {\"category\": \"car\", \"corners_3d\": [[419.39, 227.84], [372.33, 227.88], [402.49, 220.65], [443.34, 220.62], [419.39, 185.87], [372.33, 185.88], [402.49, 184.17], [443.34, 184.16]]}, {\"category\": \"car\", \"corners_3d\": [[460.79, 216.4], [429.24, 216.35], [451.77, 211.48], [479.81, 211.52], [460.79, 183.08], [429.24, 183.07], [451.77, 181.93], [479.81, 181.94]]}, {\"category\": \"car\", \"corners_3d\": [[496.72, 207.84], [470.75, 207.85], [480.18, 205.33], [504.28, 205.32], [496.72, 181.6], [470.75, 181.6], [480.18, 180.97], [504.28, 180.97]]}, {\"category\": \"car\", \"corners_3d\": [[514.54, 202.21], [491.68, 202.2], [500.88, 200.03], [522.06, 200.04], [514.54, 183.24], [491.68, 183.24], [500.88, 182.47], [522.06, 182.48]]}]\n```" - } - ] - }, - { - "id": 231, - "data_source": "Mono3DRefer", - "scene_name": "001251", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[585.82, 192.34], [601.19, 192.33], [601.82, 193.3], [585.69, 193.31], [585.82, 180.24], [601.19, 180.24], [601.82, 180.61], [585.69, 180.61]]}, {\"category\": \"car\", \"corners_3d\": [[503.76, 202.75], [481.53, 202.73], [493.22, 200.39], [513.71, 200.41], [503.76, 180.4], [481.53, 180.39], [493.22, 179.8], [513.71, 179.81]]}]\n```" - } - ] - }, - { - "id": 232, - "data_source": "Mono3DRefer", - "scene_name": "001260", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1027.76, 293.38], [1150.07, 314.74], [931.0, 337.33], [829.9, 309.29], [1027.76, 175.05], [1150.07, 175.43], [931.0, 175.85], [829.9, 175.34]]}, {\"category\": \"car\", \"corners_3d\": [[1228.97, 314.03], [1101.01, 293.78], [1267.33, 279.58], [1402.19, 295.04], [1228.97, 178.85], [1101.01, 177.99], [1267.33, 177.39], [1402.19, 178.04]]}]\n```" - } - ] - }, - { - "id": 233, - "data_source": "Mono3DRefer", - "scene_name": "001267", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[792.78, 336.43], [1008.22, 337.39], [3232.06, 1219.71], [1757.37, 1178.51], [792.78, 71.77], [1008.22, 71.09], [3232.06, -553.91], [1757.37, -524.73]]}]\n```" - } - ] - }, - { - "id": 234, - "data_source": "Mono3DRefer", - "scene_name": "001267", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[190.04, 299.46], [244.92, 293.47], [258.72, 297.21], [202.66, 303.61], [190.04, 156.26], [244.92, 157.54], [258.72, 156.74], [202.66, 155.37]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[570.9, 260.27], [529.81, 257.03], [560.42, 252.99], [600.46, 255.88], [570.9, 179.83], [529.81, 179.9], [560.42, 179.99], [600.46, 179.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[483.5, 255.72], [463.85, 258.86], [433.93, 257.15], [454.35, 254.14], [483.5, 172.96], [463.85, 172.6], [433.93, 172.8], [454.35, 173.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[555.15, 241.42], [519.35, 241.96], [516.94, 239.06], [551.06, 238.57], [555.15, 171.73], [519.35, 171.64], [516.94, 172.11], [551.06, 172.19]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[601.44, 222.99], [630.38, 223.05], [630.71, 224.77], [600.57, 224.71], [601.44, 168.55], [630.38, 168.53], [630.71, 167.99], [600.57, 168.01]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[607.92, 215.02], [591.19, 214.92], [594.06, 213.99], [610.33, 214.08], [607.92, 171.38], [591.19, 171.42], [594.06, 171.7], [610.33, 171.67]]}]\n```" - } - ] - }, - { - "id": 235, - "data_source": "Mono3DRefer", - "scene_name": "001267", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[190.04, 299.46], [244.92, 293.47], [258.72, 297.21], [202.66, 303.61], [190.04, 156.26], [244.92, 157.54], [258.72, 156.74], [202.66, 155.37]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[570.9, 260.27], [529.81, 257.03], [560.42, 252.99], [600.46, 255.88], [570.9, 179.83], [529.81, 179.9], [560.42, 179.99], [600.46, 179.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[483.5, 255.72], [463.85, 258.86], [433.93, 257.15], [454.35, 254.14], [483.5, 172.96], [463.85, 172.6], [433.93, 172.8], [454.35, 173.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[555.15, 241.42], [519.35, 241.96], [516.94, 239.06], [551.06, 238.57], [555.15, 171.73], [519.35, 171.64], [516.94, 172.11], [551.06, 172.19]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[601.44, 222.99], [630.38, 223.05], [630.71, 224.77], [600.57, 224.71], [601.44, 168.55], [630.38, 168.53], [630.71, 167.99], [600.57, 168.01]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[607.92, 215.02], [591.19, 214.92], [594.06, 213.99], [610.33, 214.08], [607.92, 171.38], [591.19, 171.42], [594.06, 171.7], [610.33, 171.67]]}, {\"category\": \"van\", \"corners_3d\": [[792.78, 336.43], [1008.22, 337.39], [3232.06, 1219.71], [1757.37, 1178.51], [792.78, 71.77], [1008.22, 71.09], [3232.06, -553.91], [1757.37, -524.73]]}]\n```" - } - ] - }, - { - "id": 236, - "data_source": "Mono3DRefer", - "scene_name": "001268", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[330.1, 304.96], [456.11, 305.45], [342.51, 394.86], [132.74, 393.51], [330.1, 188.65], [456.11, 188.71], [342.51, 199.4], [132.74, 199.24]]}, {\"category\": \"car\", \"corners_3d\": [[549.23, 222.78], [502.99, 222.49], [526.8, 215.58], [566.69, 215.8], [549.23, 173.71], [502.99, 173.7], [526.8, 173.59], [566.69, 173.59]]}, {\"category\": \"car\", \"corners_3d\": [[693.73, 247.83], [766.01, 248.06], [790.64, 261.86], [705.06, 261.54], [693.73, 176.0], [766.01, 176.01], [790.64, 176.59], [705.06, 176.58]]}, {\"category\": \"car\", \"corners_3d\": [[678.81, 231.79], [733.14, 232.06], [747.37, 242.64], [683.31, 242.26], [678.81, 185.38], [733.14, 185.43], [747.37, 187.68], [683.31, 187.6]]}, {\"category\": \"car\", \"corners_3d\": [[665.88, 199.57], [642.61, 199.54], [641.78, 197.69], [663.43, 197.71], [665.88, 178.05], [642.61, 178.04], [641.78, 177.68], [663.43, 177.69]]}]\n```" - } - ] - }, - { - "id": 237, - "data_source": "Mono3DRefer", - "scene_name": "001268", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[614.62, 209.62], [603.81, 209.61], [604.65, 208.45], [615.12, 208.47], [614.62, 174.03], [603.81, 174.03], [604.65, 173.99], [615.12, 173.99]]}]\n```" - } - ] - }, - { - "id": 238, - "data_source": "Mono3DRefer", - "scene_name": "001268", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[330.1, 304.96], [456.11, 305.45], [342.51, 394.86], [132.74, 393.51], [330.1, 188.65], [456.11, 188.71], [342.51, 199.4], [132.74, 199.24]]}, {\"category\": \"car\", \"corners_3d\": [[549.23, 222.78], [502.99, 222.49], [526.8, 215.58], [566.69, 215.8], [549.23, 173.71], [502.99, 173.7], [526.8, 173.59], [566.69, 173.59]]}, {\"category\": \"car\", \"corners_3d\": [[693.73, 247.83], [766.01, 248.06], [790.64, 261.86], [705.06, 261.54], [693.73, 176.0], [766.01, 176.01], [790.64, 176.59], [705.06, 176.58]]}, {\"category\": \"cyclist\", \"corners_3d\": [[614.62, 209.62], [603.81, 209.61], [604.65, 208.45], [615.12, 208.47], [614.62, 174.03], [603.81, 174.03], [604.65, 173.99], [615.12, 173.99]]}, {\"category\": \"car\", \"corners_3d\": [[678.81, 231.79], [733.14, 232.06], [747.37, 242.64], [683.31, 242.26], [678.81, 185.38], [733.14, 185.43], [747.37, 187.68], [683.31, 187.6]]}, {\"category\": \"car\", \"corners_3d\": [[665.88, 199.57], [642.61, 199.54], [641.78, 197.69], [663.43, 197.71], [665.88, 178.05], [642.61, 178.04], [641.78, 177.68], [663.43, 177.69]]}]\n```" - } - ] - }, - { - "id": 239, - "data_source": "Mono3DRefer", - "scene_name": "001271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[384.74, 297.45], [252.56, 298.15], [359.49, 258.13], [449.21, 257.81], [384.74, 176.85], [252.56, 176.87], [359.49, 175.59], [449.21, 175.58]]}, {\"category\": \"car\", \"corners_3d\": [[638.05, 204.77], [657.96, 204.75], [663.09, 207.38], [641.54, 207.4], [638.05, 186.98], [657.96, 186.97], [663.09, 188.14], [641.54, 188.15]]}]\n```" - } - ] - }, - { - "id": 240, - "data_source": "Mono3DRefer", - "scene_name": "001272", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[18.37, 301.67], [43.3, 292.41], [151.25, 287.56], [136.0, 296.0], [18.37, 142.66], [43.3, 145.55], [151.25, 147.07], [136.0, 144.43]]}]\n```" - } - ] - }, - { - "id": 241, - "data_source": "Mono3DRefer", - "scene_name": "001303", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[387.65, 236.67], [341.93, 234.17], [442.26, 224.59], [484.03, 226.36], [387.65, 182.12], [341.93, 181.75], [442.26, 180.37], [484.03, 180.62]]}, {\"category\": \"car\", \"corners_3d\": [[592.22, 216.18], [566.52, 213.63], [668.62, 210.13], [697.8, 212.25], [592.22, 173.98], [566.52, 173.92], [668.62, 173.83], [697.8, 173.88]]}, {\"category\": \"car\", \"corners_3d\": [[842.06, 227.66], [866.68, 232.82], [697.36, 233.39], [687.22, 228.14], [842.06, 167.69], [866.68, 167.2], [697.36, 167.15], [687.22, 167.65]]}]\n```" - } - ] - }, - { - "id": 242, - "data_source": "Mono3DRefer", - "scene_name": "001304", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[645.68, 189.36], [669.27, 189.32], [691.9, 192.32], [664.08, 192.38], [645.68, 163.51], [669.27, 163.53], [691.9, 161.83], [664.08, 161.8]]}]\n```" - } - ] - }, - { - "id": 243, - "data_source": "Mono3DRefer", - "scene_name": "001304", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[601.21, 200.89], [628.78, 200.9], [630.13, 202.92], [600.57, 202.92], [601.21, 174.44], [628.78, 174.44], [630.13, 174.56], [600.57, 174.56]]}, {\"category\": \"car\", \"corners_3d\": [[121.71, 248.02], [71.78, 248.03], [143.95, 237.93], [187.17, 237.93], [121.71, 202.28], [71.78, 202.28], [143.95, 198.32], [187.17, 198.32]]}, {\"category\": \"car\", \"corners_3d\": [[440.03, 206.49], [416.85, 206.48], [429.5, 204.35], [451.21, 204.36], [440.03, 185.97], [416.85, 185.96], [429.5, 185.13], [451.21, 185.13]]}]\n```" - } - ] - }, - { - "id": 244, - "data_source": "Mono3DRefer", - "scene_name": "001304", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[601.21, 200.89], [628.78, 200.9], [630.13, 202.92], [600.57, 202.92], [601.21, 174.44], [628.78, 174.44], [630.13, 174.56], [600.57, 174.56]]}, {\"category\": \"bus\", \"corners_3d\": [[645.68, 189.36], [669.27, 189.32], [691.9, 192.32], [664.08, 192.38], [645.68, 163.51], [669.27, 163.53], [691.9, 161.83], [664.08, 161.8]]}, {\"category\": \"car\", \"corners_3d\": [[121.71, 248.02], [71.78, 248.03], [143.95, 237.93], [187.17, 237.93], [121.71, 202.28], [71.78, 202.28], [143.95, 198.32], [187.17, 198.32]]}, {\"category\": \"car\", \"corners_3d\": [[440.03, 206.49], [416.85, 206.48], [429.5, 204.35], [451.21, 204.36], [440.03, 185.97], [416.85, 185.96], [429.5, 185.13], [451.21, 185.13]]}]\n```" - } - ] - }, - { - "id": 245, - "data_source": "Mono3DRefer", - "scene_name": "001350", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[583.93, 186.62], [604.51, 186.61], [605.6, 187.54], [583.62, 187.55], [583.93, 170.4], [604.51, 170.41], [605.6, 170.24], [583.62, 170.24]]}, {\"category\": \"car\", \"corners_3d\": [[559.78, 184.25], [577.33, 184.25], [576.7, 184.89], [558.16, 184.9], [559.78, 169.51], [577.33, 169.52], [576.7, 169.33], [558.16, 169.33]]}]\n```" - } - ] - }, - { - "id": 246, - "data_source": "Mono3DRefer", - "scene_name": "001432", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[651.25, 219.69], [688.22, 219.67], [702.27, 227.38], [659.21, 227.41], [651.25, 186.66], [688.22, 186.66], [702.27, 188.93], [659.21, 188.94]]}]\n```" - } - ] - }, - { - "id": 247, - "data_source": "Mono3DRefer", - "scene_name": "001435", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[668.84, 221.37], [713.08, 221.72], [715.51, 229.18], [664.56, 228.7], [668.84, 178.09], [713.08, 178.13], [715.51, 178.93], [664.56, 178.88]]}, {\"category\": \"car\", \"corners_3d\": [[-278.74, 394.51], [-454.53, 387.67], [39.36, 294.86], [146.67, 297.04], [-278.74, 196.99], [-454.53, 196.25], [39.36, 186.14], [146.67, 186.38]]}, {\"category\": \"car\", \"corners_3d\": [[434.53, 234.67], [376.95, 234.11], [432.56, 223.18], [480.2, 223.56], [434.53, 184.63], [376.95, 184.53], [432.56, 182.45], [480.2, 182.52]]}, {\"category\": \"car\", \"corners_3d\": [[523.8, 217.9], [485.95, 217.59], [509.93, 212.65], [543.72, 212.9], [523.8, 182.96], [485.95, 182.89], [509.93, 181.79], [543.72, 181.84]]}, {\"category\": \"car\", \"corners_3d\": [[555.06, 209.65], [528.15, 209.5], [543.75, 205.94], [568.1, 206.06], [555.06, 180.92], [528.15, 180.89], [543.75, 180.11], [568.1, 180.13]]}, {\"category\": \"car\", \"corners_3d\": [[581.02, 203.46], [557.72, 203.34], [566.77, 201.35], [588.58, 201.45], [581.02, 179.48], [557.72, 179.46], [566.77, 179.03], [588.58, 179.05]]}]\n```" - } - ] - }, - { - "id": 248, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[-3271.24, 1046.24], [-4661.93, 1063.42], [230.35, 235.7], [323.49, 235.61], [-3271.24, -488.38], [-4661.93, -501.39], [230.35, 125.28], [323.49, 125.34]]}, {\"category\": \"tram\", \"corners_3d\": [[324.03, 235.85], [229.98, 235.86], [404.62, 206.83], [455.32, 206.83], [324.03, 122.43], [229.98, 122.43], [404.62, 145.66], [455.32, 145.66]]}]\n```" - } - ] - }, - { - "id": 249, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[825.16, 227.78], [843.91, 227.76], [864.2, 232.38], [843.88, 232.4], [825.16, 163.25], [843.91, 163.26], [864.2, 162.45], [843.88, 162.45]]}]\n```" - } - ] - }, - { - "id": 250, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[458.19, 196.39], [439.78, 196.39], [449.07, 195.1], [466.48, 195.1], [458.19, 180.8], [439.78, 180.8], [449.07, 180.36], [466.48, 180.36]]}]\n```" - } - ] - }, - { - "id": 251, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[825.16, 227.78], [843.91, 227.76], [864.2, 232.38], [843.88, 232.4], [825.16, 163.25], [843.91, 163.26], [864.2, 162.45], [843.88, 162.45]]}, {\"category\": \"tram\", \"corners_3d\": [[-3271.24, 1046.24], [-4661.93, 1063.42], [230.35, 235.7], [323.49, 235.61], [-3271.24, -488.38], [-4661.93, -501.39], [230.35, 125.28], [323.49, 125.34]]}, {\"category\": \"tram\", \"corners_3d\": [[324.03, 235.85], [229.98, 235.86], [404.62, 206.83], [455.32, 206.83], [324.03, 122.43], [229.98, 122.43], [404.62, 145.66], [455.32, 145.66]]}]\n```" - } - ] - }, - { - "id": 252, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[458.19, 196.39], [439.78, 196.39], [449.07, 195.1], [466.48, 195.1], [458.19, 180.8], [439.78, 180.8], [449.07, 180.36], [466.48, 180.36]]}, {\"category\": \"tram\", \"corners_3d\": [[-3271.24, 1046.24], [-4661.93, 1063.42], [230.35, 235.7], [323.49, 235.61], [-3271.24, -488.38], [-4661.93, -501.39], [230.35, 125.28], [323.49, 125.34]]}, {\"category\": \"tram\", \"corners_3d\": [[324.03, 235.85], [229.98, 235.86], [404.62, 206.83], [455.32, 206.83], [324.03, 122.43], [229.98, 122.43], [404.62, 145.66], [455.32, 145.66]]}]\n```" - } - ] - }, - { - "id": 253, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[825.16, 227.78], [843.91, 227.76], [864.2, 232.38], [843.88, 232.4], [825.16, 163.25], [843.91, 163.26], [864.2, 162.45], [843.88, 162.45]]}, {\"category\": \"car\", \"corners_3d\": [[458.19, 196.39], [439.78, 196.39], [449.07, 195.1], [466.48, 195.1], [458.19, 180.8], [439.78, 180.8], [449.07, 180.36], [466.48, 180.36]]}]\n```" - } - ] - }, - { - "id": 254, - "data_source": "Mono3DRefer", - "scene_name": "001450", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[825.16, 227.78], [843.91, 227.76], [864.2, 232.38], [843.88, 232.4], [825.16, 163.25], [843.91, 163.26], [864.2, 162.45], [843.88, 162.45]]}, {\"category\": \"car\", \"corners_3d\": [[458.19, 196.39], [439.78, 196.39], [449.07, 195.1], [466.48, 195.1], [458.19, 180.8], [439.78, 180.8], [449.07, 180.36], [466.48, 180.36]]}, {\"category\": \"tram\", \"corners_3d\": [[-3271.24, 1046.24], [-4661.93, 1063.42], [230.35, 235.7], [323.49, 235.61], [-3271.24, -488.38], [-4661.93, -501.39], [230.35, 125.28], [323.49, 125.34]]}, {\"category\": \"tram\", \"corners_3d\": [[324.03, 235.85], [229.98, 235.86], [404.62, 206.83], [455.32, 206.83], [324.03, 122.43], [229.98, 122.43], [404.62, 145.66], [455.32, 145.66]]}]\n```" - } - ] - }, - { - "id": 255, - "data_source": "Mono3DRefer", - "scene_name": "001478", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[712.61, 228.22], [777.13, 228.42], [809.58, 241.48], [729.82, 241.18], [712.61, 174.4], [777.13, 174.41], [809.58, 174.77], [729.82, 174.76]]}]\n```" - } - ] - }, - { - "id": 256, - "data_source": "Mono3DRefer", - "scene_name": "001478", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[769.41, 185.78], [796.64, 185.99], [779.4, 187.32], [749.78, 187.05], [769.41, 150.53], [796.64, 150.16], [779.4, 147.87], [749.78, 148.33]]}]\n```" - } - ] - }, - { - "id": 257, - "data_source": "Mono3DRefer", - "scene_name": "001478", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[712.61, 228.22], [777.13, 228.42], [809.58, 241.48], [729.82, 241.18], [712.61, 174.4], [777.13, 174.41], [809.58, 174.77], [729.82, 174.76]]}, {\"category\": \"van\", \"corners_3d\": [[769.41, 185.78], [796.64, 185.99], [779.4, 187.32], [749.78, 187.05], [769.41, 150.53], [796.64, 150.16], [779.4, 147.87], [749.78, 148.33]]}]\n```" - } - ] - }, - { - "id": 258, - "data_source": "Mono3DRefer", - "scene_name": "001480", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[659.38, 210.27], [718.68, 210.27], [786.81, 233.76], [690.28, 233.75], [659.38, 146.02], [718.68, 146.02], [786.81, 129.17], [690.28, 129.18]]}]\n```" - } - ] - }, - { - "id": 259, - "data_source": "Mono3DRefer", - "scene_name": "001480", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[370.89, 215.62], [338.32, 215.56], [363.52, 211.88], [393.31, 211.93], [370.89, 187.11], [338.32, 187.09], [363.52, 185.86], [393.31, 185.87]]}, {\"category\": \"car\", \"corners_3d\": [[462.05, 196.68], [443.33, 196.67], [453.61, 195.31], [471.27, 195.32], [462.05, 181.32], [443.33, 181.32], [453.61, 180.83], [471.27, 180.84]]}]\n```" - } - ] - }, - { - "id": 260, - "data_source": "Mono3DRefer", - "scene_name": "001480", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[659.38, 210.27], [718.68, 210.27], [786.81, 233.76], [690.28, 233.75], [659.38, 146.02], [718.68, 146.02], [786.81, 129.17], [690.28, 129.18]]}, {\"category\": \"car\", \"corners_3d\": [[370.89, 215.62], [338.32, 215.56], [363.52, 211.88], [393.31, 211.93], [370.89, 187.11], [338.32, 187.09], [363.52, 185.86], [393.31, 185.87]]}, {\"category\": \"car\", \"corners_3d\": [[462.05, 196.68], [443.33, 196.67], [453.61, 195.31], [471.27, 195.32], [462.05, 181.32], [443.33, 181.32], [453.61, 180.83], [471.27, 180.84]]}]\n```" - } - ] - }, - { - "id": 261, - "data_source": "Mono3DRefer", - "scene_name": "001497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[575.56, 280.24], [678.48, 280.25], [713.91, 335.94], [557.64, 335.91], [575.56, 184.41], [678.48, 184.41], [713.91, 190.4], [557.64, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[222.04, 329.4], [378.58, 329.43], [115.97, 507.01], [-217.95, 506.88], [222.04, 178.51], [378.58, 178.51], [115.97, 184.94], [-217.95, 184.93]]}, {\"category\": \"car\", \"corners_3d\": [[467.87, 226.63], [523.86, 226.68], [502.17, 239.17], [433.23, 239.1], [467.87, 179.32], [523.86, 179.33], [502.17, 180.83], [433.23, 180.82]]}, {\"category\": \"car\", \"corners_3d\": [[691.01, 211.02], [729.19, 211.02], [747.49, 216.89], [703.44, 216.89], [691.01, 178.37], [729.19, 178.37], [747.49, 179.21], [703.44, 179.21]]}]\n```" - } - ] - }, - { - "id": 262, - "data_source": "Mono3DRefer", - "scene_name": "001497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[408.72, 247.09], [503.66, 247.1], [448.37, 285.66], [304.13, 285.64], [408.72, 150.34], [503.66, 150.33], [448.37, 138.64], [304.13, 138.65]]}, {\"category\": \"van\", \"corners_3d\": [[493.17, 213.21], [536.38, 213.16], [526.67, 219.76], [476.35, 219.83], [493.17, 168.81], [536.38, 168.81], [526.67, 168.15], [476.35, 168.15]]}, {\"category\": \"van\", \"corners_3d\": [[669.24, 200.95], [706.23, 200.95], [718.88, 204.65], [677.02, 204.65], [669.24, 166.32], [706.23, 166.32], [718.88, 165.46], [677.02, 165.46]]}]\n```" - } - ] - }, - { - "id": 263, - "data_source": "Mono3DRefer", - "scene_name": "001497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[575.56, 280.24], [678.48, 280.25], [713.91, 335.94], [557.64, 335.91], [575.56, 184.41], [678.48, 184.41], [713.91, 190.4], [557.64, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[222.04, 329.4], [378.58, 329.43], [115.97, 507.01], [-217.95, 506.88], [222.04, 178.51], [378.58, 178.51], [115.97, 184.94], [-217.95, 184.93]]}, {\"category\": \"van\", \"corners_3d\": [[408.72, 247.09], [503.66, 247.1], [448.37, 285.66], [304.13, 285.64], [408.72, 150.34], [503.66, 150.33], [448.37, 138.64], [304.13, 138.65]]}, {\"category\": \"car\", \"corners_3d\": [[467.87, 226.63], [523.86, 226.68], [502.17, 239.17], [433.23, 239.1], [467.87, 179.32], [523.86, 179.33], [502.17, 180.83], [433.23, 180.82]]}, {\"category\": \"van\", \"corners_3d\": [[493.17, 213.21], [536.38, 213.16], [526.67, 219.76], [476.35, 219.83], [493.17, 168.81], [536.38, 168.81], [526.67, 168.15], [476.35, 168.15]]}, {\"category\": \"car\", \"corners_3d\": [[691.01, 211.02], [729.19, 211.02], [747.49, 216.89], [703.44, 216.89], [691.01, 178.37], [729.19, 178.37], [747.49, 179.21], [703.44, 179.21]]}, {\"category\": \"van\", \"corners_3d\": [[669.24, 200.95], [706.23, 200.95], [718.88, 204.65], [677.02, 204.65], [669.24, 166.32], [706.23, 166.32], [718.88, 165.46], [677.02, 165.46]]}]\n```" - } - ] - }, - { - "id": 264, - "data_source": "Mono3DRefer", - "scene_name": "001502", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[680.32, 163.09], [653.01, 163.11], [652.95, 163.97], [677.83, 163.96], [680.32, 137.84], [653.01, 137.92], [652.95, 141.02], [677.83, 140.95]]}]\n```" - } - ] - }, - { - "id": 265, - "data_source": "Mono3DRefer", - "scene_name": "001502", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (motorcyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"motorcyclist\", \"corners_3d\": [[340.52, 398.07], [240.9, 396.83], [335.82, 343.28], [411.91, 343.99], [340.52, 153.91], [240.9, 154.01], [335.82, 158.52], [411.91, 158.46]]}]\n```" - } - ] - }, - { - "id": 266, - "data_source": "Mono3DRefer", - "scene_name": "001502", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, motorcyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"motorcyclist\", \"corners_3d\": [[340.52, 398.07], [240.9, 396.83], [335.82, 343.28], [411.91, 343.99], [340.52, 153.91], [240.9, 154.01], [335.82, 158.52], [411.91, 158.46]]}, {\"category\": \"car\", \"corners_3d\": [[680.32, 163.09], [653.01, 163.11], [652.95, 163.97], [677.83, 163.96], [680.32, 137.84], [653.01, 137.92], [652.95, 141.02], [677.83, 140.95]]}]\n```" - } - ] - }, - { - "id": 267, - "data_source": "Mono3DRefer", - "scene_name": "001555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[455.21, 256.56], [379.85, 256.56], [427.38, 239.2], [487.11, 239.19], [455.21, 192.37], [379.85, 192.37], [427.38, 188.32], [487.11, 188.32]]}, {\"category\": \"car\", \"corners_3d\": [[522.06, 212.84], [486.86, 212.69], [506.84, 208.26], [538.19, 208.39], [522.06, 177.99], [486.86, 177.97], [506.84, 177.4], [538.19, 177.42]]}, {\"category\": \"car\", \"corners_3d\": [[614.61, 218.16], [658.31, 218.24], [661.32, 223.38], [612.67, 223.28], [614.61, 174.24], [658.31, 174.25], [661.32, 174.4], [612.67, 174.4]]}, {\"category\": \"car\", \"corners_3d\": [[19.46, 521.66], [-319.8, 531.17], [163.69, 335.51], [314.05, 333.53], [19.46, 232.25], [-319.8, 233.87], [163.69, 200.55], [314.05, 200.21]]}]\n```" - } - ] - }, - { - "id": 268, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[1003.6, 320.67], [1035.49, 320.61], [1115.11, 347.81], [1077.39, 347.9], [1003.6, 192.42], [1035.49, 192.41], [1115.11, 196.01], [1077.39, 196.02]]}]\n```" - } - ] - }, - { - "id": 269, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[701.39, 257.04], [780.43, 256.86], [852.21, 289.51], [742.58, 289.85], [701.39, 173.79], [780.43, 173.79], [852.21, 174.16], [742.58, 174.16]]}, {\"category\": \"van\", \"corners_3d\": [[664.54, 218.8], [706.92, 218.72], [728.51, 227.09], [678.43, 227.2], [664.54, 170.43], [706.92, 170.43], [728.51, 169.99], [678.43, 169.98]]}]\n```" - } - ] - }, - { - "id": 270, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-344.77, 1129.37], [-1277.52, 1154.4], [77.18, 444.11], [330.08, 442.17], [-344.77, 321.97], [-1277.52, 325.87], [77.18, 215.14], [330.08, 214.84]]}, {\"category\": \"car\", \"corners_3d\": [[894.52, 427.28], [1152.18, 427.35], [2722.35, 1164.55], [1717.41, 1163.46], [894.52, 178.86], [1152.18, 178.86], [2722.35, 196.26], [1717.41, 196.23]]}, {\"category\": \"car\", \"corners_3d\": [[761.16, 303.9], [877.61, 303.91], [1029.57, 378.37], [846.93, 378.33], [761.16, 198.27], [877.61, 198.27], [1029.57, 212.71], [846.93, 212.71]]}, {\"category\": \"car\", \"corners_3d\": [[563.05, 228.07], [514.47, 228.08], [528.69, 219.77], [569.96, 219.76], [563.05, 183.25], [514.47, 183.25], [528.69, 181.69], [569.96, 181.69]]}, {\"category\": \"car\", \"corners_3d\": [[656.14, 214.27], [685.83, 214.24], [697.26, 219.49], [663.81, 219.53], [656.14, 183.16], [685.83, 183.15], [697.26, 184.45], [663.81, 184.46]]}]\n```" - } - ] - }, - { - "id": 271, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[1003.6, 320.67], [1035.49, 320.61], [1115.11, 347.81], [1077.39, 347.9], [1003.6, 192.42], [1035.49, 192.41], [1115.11, 196.01], [1077.39, 196.02]]}, {\"category\": \"van\", \"corners_3d\": [[701.39, 257.04], [780.43, 256.86], [852.21, 289.51], [742.58, 289.85], [701.39, 173.79], [780.43, 173.79], [852.21, 174.16], [742.58, 174.16]]}, {\"category\": \"van\", \"corners_3d\": [[664.54, 218.8], [706.92, 218.72], [728.51, 227.09], [678.43, 227.2], [664.54, 170.43], [706.92, 170.43], [728.51, 169.99], [678.43, 169.98]]}]\n```" - } - ] - }, - { - "id": 272, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[1003.6, 320.67], [1035.49, 320.61], [1115.11, 347.81], [1077.39, 347.9], [1003.6, 192.42], [1035.49, 192.41], [1115.11, 196.01], [1077.39, 196.02]]}, {\"category\": \"car\", \"corners_3d\": [[-344.77, 1129.37], [-1277.52, 1154.4], [77.18, 444.11], [330.08, 442.17], [-344.77, 321.97], [-1277.52, 325.87], [77.18, 215.14], [330.08, 214.84]]}, {\"category\": \"car\", \"corners_3d\": [[894.52, 427.28], [1152.18, 427.35], [2722.35, 1164.55], [1717.41, 1163.46], [894.52, 178.86], [1152.18, 178.86], [2722.35, 196.26], [1717.41, 196.23]]}, {\"category\": \"car\", \"corners_3d\": [[761.16, 303.9], [877.61, 303.91], [1029.57, 378.37], [846.93, 378.33], [761.16, 198.27], [877.61, 198.27], [1029.57, 212.71], [846.93, 212.71]]}, {\"category\": \"car\", \"corners_3d\": [[563.05, 228.07], [514.47, 228.08], [528.69, 219.77], [569.96, 219.76], [563.05, 183.25], [514.47, 183.25], [528.69, 181.69], [569.96, 181.69]]}, {\"category\": \"car\", \"corners_3d\": [[656.14, 214.27], [685.83, 214.24], [697.26, 219.49], [663.81, 219.53], [656.14, 183.16], [685.83, 183.15], [697.26, 184.45], [663.81, 184.46]]}]\n```" - } - ] - }, - { - "id": 273, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-344.77, 1129.37], [-1277.52, 1154.4], [77.18, 444.11], [330.08, 442.17], [-344.77, 321.97], [-1277.52, 325.87], [77.18, 215.14], [330.08, 214.84]]}, {\"category\": \"car\", \"corners_3d\": [[894.52, 427.28], [1152.18, 427.35], [2722.35, 1164.55], [1717.41, 1163.46], [894.52, 178.86], [1152.18, 178.86], [2722.35, 196.26], [1717.41, 196.23]]}, {\"category\": \"car\", \"corners_3d\": [[761.16, 303.9], [877.61, 303.91], [1029.57, 378.37], [846.93, 378.33], [761.16, 198.27], [877.61, 198.27], [1029.57, 212.71], [846.93, 212.71]]}, {\"category\": \"van\", \"corners_3d\": [[701.39, 257.04], [780.43, 256.86], [852.21, 289.51], [742.58, 289.85], [701.39, 173.79], [780.43, 173.79], [852.21, 174.16], [742.58, 174.16]]}, {\"category\": \"car\", \"corners_3d\": [[563.05, 228.07], [514.47, 228.08], [528.69, 219.77], [569.96, 219.76], [563.05, 183.25], [514.47, 183.25], [528.69, 181.69], [569.96, 181.69]]}, {\"category\": \"van\", \"corners_3d\": [[664.54, 218.8], [706.92, 218.72], [728.51, 227.09], [678.43, 227.2], [664.54, 170.43], [706.92, 170.43], [728.51, 169.99], [678.43, 169.98]]}, {\"category\": \"car\", \"corners_3d\": [[656.14, 214.27], [685.83, 214.24], [697.26, 219.49], [663.81, 219.53], [656.14, 183.16], [685.83, 183.15], [697.26, 184.45], [663.81, 184.46]]}]\n```" - } - ] - }, - { - "id": 274, - "data_source": "Mono3DRefer", - "scene_name": "001557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[1003.6, 320.67], [1035.49, 320.61], [1115.11, 347.81], [1077.39, 347.9], [1003.6, 192.42], [1035.49, 192.41], [1115.11, 196.01], [1077.39, 196.02]]}, {\"category\": \"car\", \"corners_3d\": [[-344.77, 1129.37], [-1277.52, 1154.4], [77.18, 444.11], [330.08, 442.17], [-344.77, 321.97], [-1277.52, 325.87], [77.18, 215.14], [330.08, 214.84]]}, {\"category\": \"car\", \"corners_3d\": [[894.52, 427.28], [1152.18, 427.35], [2722.35, 1164.55], [1717.41, 1163.46], [894.52, 178.86], [1152.18, 178.86], [2722.35, 196.26], [1717.41, 196.23]]}, {\"category\": \"car\", \"corners_3d\": [[761.16, 303.9], [877.61, 303.91], [1029.57, 378.37], [846.93, 378.33], [761.16, 198.27], [877.61, 198.27], [1029.57, 212.71], [846.93, 212.71]]}, {\"category\": \"van\", \"corners_3d\": [[701.39, 257.04], [780.43, 256.86], [852.21, 289.51], [742.58, 289.85], [701.39, 173.79], [780.43, 173.79], [852.21, 174.16], [742.58, 174.16]]}, {\"category\": \"car\", \"corners_3d\": [[563.05, 228.07], [514.47, 228.08], [528.69, 219.77], [569.96, 219.76], [563.05, 183.25], [514.47, 183.25], [528.69, 181.69], [569.96, 181.69]]}, {\"category\": \"van\", \"corners_3d\": [[664.54, 218.8], [706.92, 218.72], [728.51, 227.09], [678.43, 227.2], [664.54, 170.43], [706.92, 170.43], [728.51, 169.99], [678.43, 169.98]]}, {\"category\": \"car\", \"corners_3d\": [[656.14, 214.27], [685.83, 214.24], [697.26, 219.49], [663.81, 219.53], [656.14, 183.16], [685.83, 183.15], [697.26, 184.45], [663.81, 184.46]]}]\n```" - } - ] - }, - { - "id": 275, - "data_source": "Mono3DRefer", - "scene_name": "001645", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.69, 253.58], [692.61, 254.01], [668.24, 237.49], [727.4, 237.22], [766.69, 177.09], [692.61, 177.11], [668.24, 176.24], [727.4, 176.23]]}, {\"category\": \"car\", \"corners_3d\": [[516.77, 232.28], [460.69, 232.28], [486.03, 222.12], [532.53, 222.12], [516.77, 183.59], [460.69, 183.59], [486.03, 181.75], [532.53, 181.75]]}, {\"category\": \"car\", \"corners_3d\": [[646.39, 223.67], [690.59, 223.46], [711.91, 231.69], [660.6, 231.98], [646.39, 182.39], [690.59, 182.35], [711.91, 183.9], [660.6, 183.95]]}, {\"category\": \"car\", \"corners_3d\": [[205.81, 432.25], [-59.05, 436.96], [243.11, 310.68], [379.68, 309.38], [205.81, 208.32], [-59.05, 208.96], [243.11, 191.7], [379.68, 191.52]]}]\n```" - } - ] - }, - { - "id": 276, - "data_source": "Mono3DRefer", - "scene_name": "001650", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[658.55, 215.39], [688.52, 215.37], [699.61, 220.88], [665.76, 220.9], [658.55, 188.61], [688.52, 188.6], [699.61, 190.64], [665.76, 190.65]]}]\n```" - } - ] - }, - { - "id": 277, - "data_source": "Mono3DRefer", - "scene_name": "001660", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[144.66, 236.85], [99.27, 236.82], [168.13, 228.3], [207.5, 228.33], [144.66, 194.93], [99.27, 194.92], [168.13, 191.98], [207.5, 191.99]]}, {\"category\": \"car\", \"corners_3d\": [[330.27, 217.51], [296.33, 217.47], [322.77, 213.86], [353.98, 213.9], [330.27, 187.67], [296.33, 187.66], [322.77, 186.46], [353.98, 186.47]]}]\n```" - } - ] - }, - { - "id": 278, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[497.43, 248.08], [421.48, 247.69], [479.18, 228.38], [535.68, 228.59], [497.43, 131.64], [421.48, 131.85], [479.18, 142.43], [535.68, 142.32]]}]\n```" - } - ] - }, - { - "id": 279, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[599.6, 233.35], [612.51, 232.54], [620.83, 233.44], [607.83, 234.28], [599.6, 178.7], [612.51, 178.62], [620.83, 178.71], [607.83, 178.79]]}]\n```" - } - ] - }, - { - "id": 280, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[176.93, 450.73], [-74.84, 451.77], [230.33, 325.98], [368.15, 325.66], [176.93, 204.52], [-74.84, 204.64], [230.33, 190.3], [368.15, 190.27]]}, {\"category\": \"car\", \"corners_3d\": [[616.54, 227.69], [662.87, 227.66], [672.07, 236.07], [618.63, 236.11], [616.54, 183.45], [662.87, 183.44], [672.07, 185.07], [618.63, 185.08]]}]\n```" - } - ] - }, - { - "id": 281, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[497.43, 248.08], [421.48, 247.69], [479.18, 228.38], [535.68, 228.59], [497.43, 131.64], [421.48, 131.85], [479.18, 142.43], [535.68, 142.32]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[599.6, 233.35], [612.51, 232.54], [620.83, 233.44], [607.83, 234.28], [599.6, 178.7], [612.51, 178.62], [620.83, 178.71], [607.83, 178.79]]}]\n```" - } - ] - }, - { - "id": 282, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[176.93, 450.73], [-74.84, 451.77], [230.33, 325.98], [368.15, 325.66], [176.93, 204.52], [-74.84, 204.64], [230.33, 190.3], [368.15, 190.27]]}, {\"category\": \"truck\", \"corners_3d\": [[497.43, 248.08], [421.48, 247.69], [479.18, 228.38], [535.68, 228.59], [497.43, 131.64], [421.48, 131.85], [479.18, 142.43], [535.68, 142.32]]}, {\"category\": \"car\", \"corners_3d\": [[616.54, 227.69], [662.87, 227.66], [672.07, 236.07], [618.63, 236.11], [616.54, 183.45], [662.87, 183.44], [672.07, 185.07], [618.63, 185.08]]}]\n```" - } - ] - }, - { - "id": 283, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[176.93, 450.73], [-74.84, 451.77], [230.33, 325.98], [368.15, 325.66], [176.93, 204.52], [-74.84, 204.64], [230.33, 190.3], [368.15, 190.27]]}, {\"category\": \"car\", \"corners_3d\": [[616.54, 227.69], [662.87, 227.66], [672.07, 236.07], [618.63, 236.11], [616.54, 183.45], [662.87, 183.44], [672.07, 185.07], [618.63, 185.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[599.6, 233.35], [612.51, 232.54], [620.83, 233.44], [607.83, 234.28], [599.6, 178.7], [612.51, 178.62], [620.83, 178.71], [607.83, 178.79]]}]\n```" - } - ] - }, - { - "id": 284, - "data_source": "Mono3DRefer", - "scene_name": "001665", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[176.93, 450.73], [-74.84, 451.77], [230.33, 325.98], [368.15, 325.66], [176.93, 204.52], [-74.84, 204.64], [230.33, 190.3], [368.15, 190.27]]}, {\"category\": \"truck\", \"corners_3d\": [[497.43, 248.08], [421.48, 247.69], [479.18, 228.38], [535.68, 228.59], [497.43, 131.64], [421.48, 131.85], [479.18, 142.43], [535.68, 142.32]]}, {\"category\": \"car\", \"corners_3d\": [[616.54, 227.69], [662.87, 227.66], [672.07, 236.07], [618.63, 236.11], [616.54, 183.45], [662.87, 183.44], [672.07, 185.07], [618.63, 185.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[599.6, 233.35], [612.51, 232.54], [620.83, 233.44], [607.83, 234.28], [599.6, 178.7], [612.51, 178.62], [620.83, 178.71], [607.83, 178.79]]}]\n```" - } - ] - }, - { - "id": 285, - "data_source": "Mono3DRefer", - "scene_name": "001680", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[727.51, 217.68], [687.63, 217.64], [679.96, 212.29], [715.07, 212.32], [727.51, 176.88], [687.63, 176.87], [679.96, 176.39], [715.07, 176.39]]}, {\"category\": \"car\", \"corners_3d\": [[586.49, 209.61], [552.51, 209.48], [565.24, 205.4], [595.46, 205.51], [586.49, 179.83], [552.51, 179.8], [565.24, 179.03], [595.46, 179.05]]}, {\"category\": \"car\", \"corners_3d\": [[535.56, 236.17], [470.77, 236.11], [498.04, 224.25], [550.69, 224.29], [535.56, 178.78], [470.77, 178.77], [498.04, 177.66], [550.69, 177.67]]}]\n```" - } - ] - }, - { - "id": 286, - "data_source": "Mono3DRefer", - "scene_name": "001684", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[589.9, 190.99], [612.94, 190.98], [615.98, 194.17], [588.88, 194.19], [589.9, 166.02], [612.94, 166.02], [615.98, 164.82], [588.88, 164.81]]}]\n```" - } - ] - }, - { - "id": 287, - "data_source": "Mono3DRefer", - "scene_name": "001684", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[181.18, 233.25], [137.95, 233.25], [185.07, 227.21], [223.98, 227.2], [181.18, 197.67], [137.95, 197.68], [185.07, 195.19], [223.98, 195.19]]}, {\"category\": \"car\", \"corners_3d\": [[960.95, 198.74], [949.89, 197.97], [996.08, 197.92], [1008.52, 198.68], [960.95, 178.97], [949.89, 178.79], [996.08, 178.78], [1008.52, 178.96]]}]\n```" - } - ] - }, - { - "id": 288, - "data_source": "Mono3DRefer", - "scene_name": "001684", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (bus, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"bus\", \"corners_3d\": [[589.9, 190.99], [612.94, 190.98], [615.98, 194.17], [588.88, 194.19], [589.9, 166.02], [612.94, 166.02], [615.98, 164.82], [588.88, 164.81]]}, {\"category\": \"car\", \"corners_3d\": [[181.18, 233.25], [137.95, 233.25], [185.07, 227.21], [223.98, 227.2], [181.18, 197.67], [137.95, 197.68], [185.07, 195.19], [223.98, 195.19]]}, {\"category\": \"car\", \"corners_3d\": [[960.95, 198.74], [949.89, 197.97], [996.08, 197.92], [1008.52, 198.68], [960.95, 178.97], [949.89, 178.79], [996.08, 178.78], [1008.52, 178.96]]}]\n```" - } - ] - }, - { - "id": 289, - "data_source": "Mono3DRefer", - "scene_name": "001694", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[524.63, 245.11], [473.98, 245.21], [492.47, 234.12], [535.34, 234.04], [524.63, 198.38], [473.98, 198.42], [492.47, 194.5], [535.34, 194.47]]}]\n```" - } - ] - }, - { - "id": 290, - "data_source": "Mono3DRefer", - "scene_name": "001710", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[329.0, 357.27], [166.92, 353.88], [340.99, 290.85], [448.06, 292.28], [329.0, 207.21], [166.92, 206.57], [340.99, 194.83], [448.06, 195.1]]}, {\"category\": \"car\", \"corners_3d\": [[463.59, 292.34], [354.91, 291.07], [439.18, 258.37], [518.22, 259.03], [463.59, 176.75], [354.91, 176.71], [439.18, 175.64], [518.22, 175.66]]}, {\"category\": \"car\", \"corners_3d\": [[716.44, 230.29], [768.43, 230.29], [799.83, 241.69], [737.53, 241.69], [716.44, 186.22], [768.43, 186.22], [799.83, 188.87], [737.53, 188.87]]}, {\"category\": \"car\", \"corners_3d\": [[547.46, 243.29], [489.85, 242.9], [517.28, 231.58], [565.66, 231.86], [547.46, 182.05], [489.85, 182.0], [517.28, 180.52], [565.66, 180.55]]}, {\"category\": \"car\", \"corners_3d\": [[696.27, 219.46], [741.77, 219.49], [762.98, 227.44], [709.71, 227.4], [696.27, 176.69], [741.77, 176.69], [762.98, 177.35], [709.71, 177.34]]}, {\"category\": \"car\", \"corners_3d\": [[566.46, 230.58], [524.17, 230.45], [537.07, 223.91], [574.58, 224.01], [566.46, 183.26], [524.17, 183.24], [537.07, 182.06], [574.58, 182.08]]}]\n```" - } - ] - }, - { - "id": 291, - "data_source": "Mono3DRefer", - "scene_name": "001710", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[1119.27, 232.52], [1188.02, 240.24], [942.23, 240.95], [901.26, 233.08], [1119.27, 126.81], [1188.02, 120.85], [942.23, 120.31], [901.26, 126.38]]}]\n```" - } - ] - }, - { - "id": 292, - "data_source": "Mono3DRefer", - "scene_name": "001710", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[329.0, 357.27], [166.92, 353.88], [340.99, 290.85], [448.06, 292.28], [329.0, 207.21], [166.92, 206.57], [340.99, 194.83], [448.06, 195.1]]}, {\"category\": \"car\", \"corners_3d\": [[463.59, 292.34], [354.91, 291.07], [439.18, 258.37], [518.22, 259.03], [463.59, 176.75], [354.91, 176.71], [439.18, 175.64], [518.22, 175.66]]}, {\"category\": \"van\", \"corners_3d\": [[1119.27, 232.52], [1188.02, 240.24], [942.23, 240.95], [901.26, 233.08], [1119.27, 126.81], [1188.02, 120.85], [942.23, 120.31], [901.26, 126.38]]}, {\"category\": \"car\", \"corners_3d\": [[716.44, 230.29], [768.43, 230.29], [799.83, 241.69], [737.53, 241.69], [716.44, 186.22], [768.43, 186.22], [799.83, 188.87], [737.53, 188.87]]}, {\"category\": \"car\", \"corners_3d\": [[547.46, 243.29], [489.85, 242.9], [517.28, 231.58], [565.66, 231.86], [547.46, 182.05], [489.85, 182.0], [517.28, 180.52], [565.66, 180.55]]}, {\"category\": \"car\", \"corners_3d\": [[696.27, 219.46], [741.77, 219.49], [762.98, 227.44], [709.71, 227.4], [696.27, 176.69], [741.77, 176.69], [762.98, 177.35], [709.71, 177.34]]}, {\"category\": \"car\", \"corners_3d\": [[566.46, 230.58], [524.17, 230.45], [537.07, 223.91], [574.58, 224.01], [566.46, 183.26], [524.17, 183.24], [537.07, 182.06], [574.58, 182.08]]}]\n```" - } - ] - }, - { - "id": 293, - "data_source": "Mono3DRefer", - "scene_name": "001711", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[292.82, 209.88], [269.74, 209.31], [329.29, 205.85], [350.99, 206.32], [292.82, 180.4], [269.74, 180.29], [329.29, 179.58], [350.99, 179.68]]}, {\"category\": \"car\", \"corners_3d\": [[435.11, 199.49], [423.85, 198.82], [472.53, 197.58], [484.44, 198.19], [435.11, 173.29], [423.85, 173.28], [472.53, 173.26], [484.44, 173.27]]}, {\"category\": \"car\", \"corners_3d\": [[380.57, 198.34], [373.6, 197.68], [426.22, 196.66], [434.24, 197.28], [380.57, 177.12], [373.6, 177.01], [426.22, 176.84], [434.24, 176.94]]}, {\"category\": \"car\", \"corners_3d\": [[524.33, 191.97], [523.87, 191.47], [584.51, 191.25], [586.56, 191.74], [524.33, 171.36], [523.87, 171.4], [584.51, 171.42], [586.56, 171.38]]}]\n```" - } - ] - }, - { - "id": 294, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[901.98, 198.46], [926.03, 198.73], [921.83, 199.91], [896.32, 199.6], [901.98, 150.07], [926.03, 149.61], [921.83, 147.62], [896.32, 148.14]]}, {\"category\": \"cyclist\", \"corners_3d\": [[385.44, 245.49], [391.11, 247.95], [302.02, 251.78], [299.47, 249.03], [385.44, 168.14], [391.11, 167.67], [302.02, 166.95], [299.47, 167.47]]}, {\"category\": \"cyclist\", \"corners_3d\": [[1269.43, 206.81], [1292.27, 207.81], [1208.75, 207.54], [1189.15, 206.56], [1269.43, 130.16], [1292.27, 128.25], [1208.75, 128.76], [1189.15, 130.63]]}, {\"category\": \"cyclist\", \"corners_3d\": [[936.36, 193.59], [946.09, 193.86], [905.01, 194.02], [896.02, 193.74], [936.36, 150.99], [946.09, 150.37], [905.01, 150.02], [896.02, 150.66]]}]\n```" - } - ] - }, - { - "id": 295, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[447.26, 248.43], [440.08, 251.54], [384.88, 251.54], [394.48, 248.43], [447.26, 157.07], [440.08, 156.0], [384.88, 156.0], [394.48, 157.07]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[197.5, 244.72], [181.74, 246.7], [148.61, 246.18], [165.23, 244.24], [197.5, 178.11], [181.74, 178.04], [148.61, 178.06], [165.23, 178.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[182.6, 246.46], [169.48, 248.29], [131.93, 248.03], [146.02, 246.21], [182.6, 181.64], [169.48, 181.67], [131.93, 181.66], [146.02, 181.63]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[121.94, 248.66], [137.03, 246.46], [173.04, 246.39], [159.17, 248.58], [121.94, 178.68], [137.03, 178.74], [173.04, 178.74], [159.17, 178.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[61.48, 262.24], [69.96, 259.47], [128.24, 257.42], [121.98, 260.04], [61.48, 172.98], [69.96, 173.24], [128.24, 173.42], [121.98, 173.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.57, 248.48], [240.73, 246.2], [275.17, 246.27], [263.19, 248.56], [227.57, 162.87], [240.73, 163.46], [275.17, 163.44], [263.19, 162.85]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[647.46, 234.38], [633.51, 236.46], [593.15, 234.68], [608.12, 232.73], [647.46, 154.11], [633.51, 153.09], [593.15, 153.97], [608.12, 154.92]]}]\n```" - } - ] - }, - { - "id": 296, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1196.63, 187.0], [1237.16, 187.45], [1120.36, 187.43], [1087.58, 186.98], [1196.63, 152.13], [1237.16, 150.14], [1120.36, 150.24], [1087.58, 152.21]]}, {\"category\": \"car\", \"corners_3d\": [[342.68, 278.46], [456.76, 278.48], [382.41, 327.73], [211.0, 327.71], [342.68, 182.32], [456.76, 182.32], [382.41, 183.23], [211.0, 183.23]]}]\n```" - } - ] - }, - { - "id": 297, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[447.26, 248.43], [440.08, 251.54], [384.88, 251.54], [394.48, 248.43], [447.26, 157.07], [440.08, 156.0], [384.88, 156.0], [394.48, 157.07]]}, {\"category\": \"cyclist\", \"corners_3d\": [[901.98, 198.46], [926.03, 198.73], [921.83, 199.91], [896.32, 199.6], [901.98, 150.07], [926.03, 149.61], [921.83, 147.62], [896.32, 148.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[197.5, 244.72], [181.74, 246.7], [148.61, 246.18], [165.23, 244.24], [197.5, 178.11], [181.74, 178.04], [148.61, 178.06], [165.23, 178.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[182.6, 246.46], [169.48, 248.29], [131.93, 248.03], [146.02, 246.21], [182.6, 181.64], [169.48, 181.67], [131.93, 181.66], [146.02, 181.63]]}, {\"category\": \"cyclist\", \"corners_3d\": [[385.44, 245.49], [391.11, 247.95], [302.02, 251.78], [299.47, 249.03], [385.44, 168.14], [391.11, 167.67], [302.02, 166.95], [299.47, 167.47]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[121.94, 248.66], [137.03, 246.46], [173.04, 246.39], [159.17, 248.58], [121.94, 178.68], [137.03, 178.74], [173.04, 178.74], [159.17, 178.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[61.48, 262.24], [69.96, 259.47], [128.24, 257.42], [121.98, 260.04], [61.48, 172.98], [69.96, 173.24], [128.24, 173.42], [121.98, 173.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.57, 248.48], [240.73, 246.2], [275.17, 246.27], [263.19, 248.56], [227.57, 162.87], [240.73, 163.46], [275.17, 163.44], [263.19, 162.85]]}, {\"category\": \"cyclist\", \"corners_3d\": [[1269.43, 206.81], [1292.27, 207.81], [1208.75, 207.54], [1189.15, 206.56], [1269.43, 130.16], [1292.27, 128.25], [1208.75, 128.76], [1189.15, 130.63]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[647.46, 234.38], [633.51, 236.46], [593.15, 234.68], [608.12, 232.73], [647.46, 154.11], [633.51, 153.09], [593.15, 153.97], [608.12, 154.92]]}, {\"category\": \"cyclist\", \"corners_3d\": [[936.36, 193.59], [946.09, 193.86], [905.01, 194.02], [896.02, 193.74], [936.36, 150.99], [946.09, 150.37], [905.01, 150.02], [896.02, 150.66]]}]\n```" - } - ] - }, - { - "id": 298, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[901.98, 198.46], [926.03, 198.73], [921.83, 199.91], [896.32, 199.6], [901.98, 150.07], [926.03, 149.61], [921.83, 147.62], [896.32, 148.14]]}, {\"category\": \"cyclist\", \"corners_3d\": [[385.44, 245.49], [391.11, 247.95], [302.02, 251.78], [299.47, 249.03], [385.44, 168.14], [391.11, 167.67], [302.02, 166.95], [299.47, 167.47]]}, {\"category\": \"car\", \"corners_3d\": [[1196.63, 187.0], [1237.16, 187.45], [1120.36, 187.43], [1087.58, 186.98], [1196.63, 152.13], [1237.16, 150.14], [1120.36, 150.24], [1087.58, 152.21]]}, {\"category\": \"car\", \"corners_3d\": [[342.68, 278.46], [456.76, 278.48], [382.41, 327.73], [211.0, 327.71], [342.68, 182.32], [456.76, 182.32], [382.41, 183.23], [211.0, 183.23]]}, {\"category\": \"cyclist\", \"corners_3d\": [[1269.43, 206.81], [1292.27, 207.81], [1208.75, 207.54], [1189.15, 206.56], [1269.43, 130.16], [1292.27, 128.25], [1208.75, 128.76], [1189.15, 130.63]]}, {\"category\": \"cyclist\", \"corners_3d\": [[936.36, 193.59], [946.09, 193.86], [905.01, 194.02], [896.02, 193.74], [936.36, 150.99], [946.09, 150.37], [905.01, 150.02], [896.02, 150.66]]}]\n```" - } - ] - }, - { - "id": 299, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[447.26, 248.43], [440.08, 251.54], [384.88, 251.54], [394.48, 248.43], [447.26, 157.07], [440.08, 156.0], [384.88, 156.0], [394.48, 157.07]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[197.5, 244.72], [181.74, 246.7], [148.61, 246.18], [165.23, 244.24], [197.5, 178.11], [181.74, 178.04], [148.61, 178.06], [165.23, 178.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[182.6, 246.46], [169.48, 248.29], [131.93, 248.03], [146.02, 246.21], [182.6, 181.64], [169.48, 181.67], [131.93, 181.66], [146.02, 181.63]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[121.94, 248.66], [137.03, 246.46], [173.04, 246.39], [159.17, 248.58], [121.94, 178.68], [137.03, 178.74], [173.04, 178.74], [159.17, 178.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[61.48, 262.24], [69.96, 259.47], [128.24, 257.42], [121.98, 260.04], [61.48, 172.98], [69.96, 173.24], [128.24, 173.42], [121.98, 173.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.57, 248.48], [240.73, 246.2], [275.17, 246.27], [263.19, 248.56], [227.57, 162.87], [240.73, 163.46], [275.17, 163.44], [263.19, 162.85]]}, {\"category\": \"car\", \"corners_3d\": [[1196.63, 187.0], [1237.16, 187.45], [1120.36, 187.43], [1087.58, 186.98], [1196.63, 152.13], [1237.16, 150.14], [1120.36, 150.24], [1087.58, 152.21]]}, {\"category\": \"car\", \"corners_3d\": [[342.68, 278.46], [456.76, 278.48], [382.41, 327.73], [211.0, 327.71], [342.68, 182.32], [456.76, 182.32], [382.41, 183.23], [211.0, 183.23]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[647.46, 234.38], [633.51, 236.46], [593.15, 234.68], [608.12, 232.73], [647.46, 154.11], [633.51, 153.09], [593.15, 153.97], [608.12, 154.92]]}]\n```" - } - ] - }, - { - "id": 300, - "data_source": "Mono3DRefer", - "scene_name": "001714", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[447.26, 248.43], [440.08, 251.54], [384.88, 251.54], [394.48, 248.43], [447.26, 157.07], [440.08, 156.0], [384.88, 156.0], [394.48, 157.07]]}, {\"category\": \"cyclist\", \"corners_3d\": [[901.98, 198.46], [926.03, 198.73], [921.83, 199.91], [896.32, 199.6], [901.98, 150.07], [926.03, 149.61], [921.83, 147.62], [896.32, 148.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[197.5, 244.72], [181.74, 246.7], [148.61, 246.18], [165.23, 244.24], [197.5, 178.11], [181.74, 178.04], [148.61, 178.06], [165.23, 178.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[182.6, 246.46], [169.48, 248.29], [131.93, 248.03], [146.02, 246.21], [182.6, 181.64], [169.48, 181.67], [131.93, 181.66], [146.02, 181.63]]}, {\"category\": \"cyclist\", \"corners_3d\": [[385.44, 245.49], [391.11, 247.95], [302.02, 251.78], [299.47, 249.03], [385.44, 168.14], [391.11, 167.67], [302.02, 166.95], [299.47, 167.47]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[121.94, 248.66], [137.03, 246.46], [173.04, 246.39], [159.17, 248.58], [121.94, 178.68], [137.03, 178.74], [173.04, 178.74], [159.17, 178.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[61.48, 262.24], [69.96, 259.47], [128.24, 257.42], [121.98, 260.04], [61.48, 172.98], [69.96, 173.24], [128.24, 173.42], [121.98, 173.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[227.57, 248.48], [240.73, 246.2], [275.17, 246.27], [263.19, 248.56], [227.57, 162.87], [240.73, 163.46], [275.17, 163.44], [263.19, 162.85]]}, {\"category\": \"car\", \"corners_3d\": [[1196.63, 187.0], [1237.16, 187.45], [1120.36, 187.43], [1087.58, 186.98], [1196.63, 152.13], [1237.16, 150.14], [1120.36, 150.24], [1087.58, 152.21]]}, {\"category\": \"car\", \"corners_3d\": [[342.68, 278.46], [456.76, 278.48], [382.41, 327.73], [211.0, 327.71], [342.68, 182.32], [456.76, 182.32], [382.41, 183.23], [211.0, 183.23]]}, {\"category\": \"cyclist\", \"corners_3d\": [[1269.43, 206.81], [1292.27, 207.81], [1208.75, 207.54], [1189.15, 206.56], [1269.43, 130.16], [1292.27, 128.25], [1208.75, 128.76], [1189.15, 130.63]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[647.46, 234.38], [633.51, 236.46], [593.15, 234.68], [608.12, 232.73], [647.46, 154.11], [633.51, 153.09], [593.15, 153.97], [608.12, 154.92]]}, {\"category\": \"cyclist\", \"corners_3d\": [[936.36, 193.59], [946.09, 193.86], [905.01, 194.02], [896.02, 193.74], [936.36, 150.99], [946.09, 150.37], [905.01, 150.02], [896.02, 150.66]]}]\n```" - } - ] - }, - { - "id": 301, - "data_source": "Mono3DRefer", - "scene_name": "001823", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[640.62, 204.19], [675.91, 204.26], [678.56, 207.07], [640.11, 206.98], [640.62, 168.79], [675.91, 168.78], [678.56, 168.41], [640.11, 168.43]]}, {\"category\": \"car\", \"corners_3d\": [[571.04, 206.43], [534.09, 206.19], [554.22, 202.39], [587.06, 202.58], [571.04, 174.2], [534.09, 174.19], [554.22, 174.04], [587.06, 174.04]]}, {\"category\": \"car\", \"corners_3d\": [[589.82, 198.18], [560.49, 198.15], [567.4, 195.66], [593.84, 195.68], [589.82, 171.18], [560.49, 171.19], [567.4, 171.35], [593.84, 171.35]]}]\n```" - } - ] - }, - { - "id": 302, - "data_source": "Mono3DRefer", - "scene_name": "001844", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-89.98, 264.7], [-99.03, 258.01], [137.91, 242.96], [160.35, 247.43], [-89.98, 184.18], [-99.03, 183.36], [137.91, 181.5], [160.35, 182.05]]}, {\"category\": \"car\", \"corners_3d\": [[373.31, 224.21], [364.11, 220.62], [509.54, 216.43], [527.79, 219.4], [373.31, 169.93], [364.11, 170.13], [509.54, 170.37], [527.79, 170.2]]}, {\"category\": \"car\", \"corners_3d\": [[944.1, 235.2], [1024.14, 240.5], [942.47, 252.99], [856.46, 245.67], [944.1, 162.22], [1024.14, 161.32], [942.47, 159.19], [856.46, 160.44]]}, {\"category\": \"car\", \"corners_3d\": [[1249.09, 194.0], [1206.02, 192.81], [1292.19, 192.32], [1339.17, 193.46], [1249.09, 149.15], [1206.02, 150.49], [1292.19, 151.03], [1339.17, 149.76]]}, {\"category\": \"car\", \"corners_3d\": [[1360.06, 217.67], [1437.98, 221.73], [1296.42, 223.49], [1228.1, 219.14], [1360.06, 161.16], [1437.98, 160.1], [1296.42, 159.64], [1228.1, 160.78]]}]\n```" - } - ] - }, - { - "id": 303, - "data_source": "Mono3DRefer", - "scene_name": "001844", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[936.53, 203.7], [909.07, 201.73], [1020.05, 201.06], [1054.21, 202.93], [936.53, 143.48], [909.07, 145.36], [1020.05, 146.0], [1054.21, 144.21]]}]\n```" - } - ] - }, - { - "id": 304, - "data_source": "Mono3DRefer", - "scene_name": "001844", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-89.98, 264.7], [-99.03, 258.01], [137.91, 242.96], [160.35, 247.43], [-89.98, 184.18], [-99.03, 183.36], [137.91, 181.5], [160.35, 182.05]]}, {\"category\": \"car\", \"corners_3d\": [[373.31, 224.21], [364.11, 220.62], [509.54, 216.43], [527.79, 219.4], [373.31, 169.93], [364.11, 170.13], [509.54, 170.37], [527.79, 170.2]]}, {\"category\": \"car\", \"corners_3d\": [[944.1, 235.2], [1024.14, 240.5], [942.47, 252.99], [856.46, 245.67], [944.1, 162.22], [1024.14, 161.32], [942.47, 159.19], [856.46, 160.44]]}, {\"category\": \"van\", \"corners_3d\": [[936.53, 203.7], [909.07, 201.73], [1020.05, 201.06], [1054.21, 202.93], [936.53, 143.48], [909.07, 145.36], [1020.05, 146.0], [1054.21, 144.21]]}, {\"category\": \"car\", \"corners_3d\": [[1249.09, 194.0], [1206.02, 192.81], [1292.19, 192.32], [1339.17, 193.46], [1249.09, 149.15], [1206.02, 150.49], [1292.19, 151.03], [1339.17, 149.76]]}, {\"category\": \"car\", \"corners_3d\": [[1360.06, 217.67], [1437.98, 221.73], [1296.42, 223.49], [1228.1, 219.14], [1360.06, 161.16], [1437.98, 160.1], [1296.42, 159.64], [1228.1, 160.78]]}]\n```" - } - ] - }, - { - "id": 305, - "data_source": "Mono3DRefer", - "scene_name": "001890", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[579.44, 195.89], [606.94, 195.86], [608.65, 197.99], [578.6, 198.02], [579.44, 174.22], [606.94, 174.22], [608.65, 174.35], [578.6, 174.35]]}, {\"category\": \"car\", \"corners_3d\": [[546.69, 192.28], [570.01, 192.27], [568.05, 193.75], [542.96, 193.76], [546.69, 172.71], [570.01, 172.71], [568.05, 172.69], [542.96, 172.69]]}]\n```" - } - ] - }, - { - "id": 306, - "data_source": "Mono3DRefer", - "scene_name": "001890", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[600.08, 182.91], [615.5, 182.91], [615.89, 183.64], [599.35, 183.64], [600.08, 161.66], [615.5, 161.66], [615.89, 160.85], [599.35, 160.85]]}]\n```" - } - ] - }, - { - "id": 307, - "data_source": "Mono3DRefer", - "scene_name": "001890", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[600.08, 182.91], [615.5, 182.91], [615.89, 183.64], [599.35, 183.64], [600.08, 161.66], [615.5, 161.66], [615.89, 160.85], [599.35, 160.85]]}, {\"category\": \"car\", \"corners_3d\": [[579.44, 195.89], [606.94, 195.86], [608.65, 197.99], [578.6, 198.02], [579.44, 174.22], [606.94, 174.22], [608.65, 174.35], [578.6, 174.35]]}, {\"category\": \"car\", \"corners_3d\": [[546.69, 192.28], [570.01, 192.27], [568.05, 193.75], [542.96, 193.76], [546.69, 172.71], [570.01, 172.71], [568.05, 172.69], [542.96, 172.69]]}]\n```" - } - ] - }, - { - "id": 308, - "data_source": "Mono3DRefer", - "scene_name": "001920", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[801.2, 216.07], [817.66, 216.13], [829.97, 219.37], [812.25, 219.3], [801.2, 163.74], [817.66, 163.73], [829.97, 163.05], [812.25, 163.06]]}]\n```" - } - ] - }, - { - "id": 309, - "data_source": "Mono3DRefer", - "scene_name": "001920", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[409.2, 206.17], [380.19, 206.1], [401.1, 203.48], [427.86, 203.53], [409.2, 182.21], [380.19, 182.19], [401.1, 181.45], [427.86, 181.47]]}]\n```" - } - ] - }, - { - "id": 310, - "data_source": "Mono3DRefer", - "scene_name": "001920", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[409.2, 206.17], [380.19, 206.1], [401.1, 203.48], [427.86, 203.53], [409.2, 182.21], [380.19, 182.19], [401.1, 181.45], [427.86, 181.47]]}, {\"category\": \"cyclist\", \"corners_3d\": [[801.2, 216.07], [817.66, 216.13], [829.97, 219.37], [812.25, 219.3], [801.2, 163.74], [817.66, 163.73], [829.97, 163.05], [812.25, 163.06]]}]\n```" - } - ] - }, - { - "id": 311, - "data_source": "Mono3DRefer", - "scene_name": "001985", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[515.2, 213.76], [541.5, 213.67], [539.38, 217.1], [510.86, 217.2], [515.2, 187.22], [541.5, 187.19], [539.38, 188.39], [510.86, 188.43]]}, {\"category\": \"car\", \"corners_3d\": [[397.95, 326.84], [259.24, 325.97], [370.92, 279.93], [468.2, 280.35], [397.95, 202.63], [259.24, 202.47], [370.92, 193.56], [468.2, 193.64]]}]\n```" - } - ] - }, - { - "id": 312, - "data_source": "Mono3DRefer", - "scene_name": "002010", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[264.22, 284.85], [159.47, 285.35], [289.21, 251.24], [361.9, 251.0], [264.22, 137.66], [159.47, 137.5], [289.21, 148.22], [361.9, 148.3]]}]\n```" - } - ] - }, - { - "id": 313, - "data_source": "Mono3DRefer", - "scene_name": "002010", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[590.68, 198.7], [606.06, 198.69], [606.57, 199.97], [590.44, 199.99], [590.68, 186.59], [606.06, 186.59], [606.57, 187.27], [590.44, 187.28]]}]\n```" - } - ] - }, - { - "id": 314, - "data_source": "Mono3DRefer", - "scene_name": "002010", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[264.22, 284.85], [159.47, 285.35], [289.21, 251.24], [361.9, 251.0], [264.22, 137.66], [159.47, 137.5], [289.21, 148.22], [361.9, 148.3]]}, {\"category\": \"car\", \"corners_3d\": [[590.68, 198.7], [606.06, 198.69], [606.57, 199.97], [590.44, 199.99], [590.68, 186.59], [606.06, 186.59], [606.57, 187.27], [590.44, 187.28]]}]\n```" - } - ] - }, - { - "id": 315, - "data_source": "Mono3DRefer", - "scene_name": "002022", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[529.93, 236.36], [468.89, 236.1], [496.85, 226.06], [548.27, 226.25], [529.93, 181.92], [468.89, 181.88], [496.85, 180.45], [548.27, 180.47]]}, {\"category\": \"car\", \"corners_3d\": [[551.62, 192.64], [572.36, 192.62], [571.31, 193.92], [549.2, 193.94], [551.62, 171.65], [572.36, 171.65], [571.31, 171.57], [549.2, 171.57]]}]\n```" - } - ] - }, - { - "id": 316, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[719.76, 252.73], [805.46, 252.48], [885.23, 277.52], [769.52, 277.98], [719.76, 159.69], [805.46, 159.77], [885.23, 152.1], [769.52, 151.96]]}]\n```" - } - ] - }, - { - "id": 317, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[375.62, 249.09], [405.66, 248.55], [403.82, 251.03], [372.65, 251.6], [375.62, 166.46], [405.66, 166.58], [403.82, 166.03], [372.65, 165.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[390.5, 246.95], [365.62, 246.89], [373.51, 244.87], [397.62, 244.93], [390.5, 165.29], [365.62, 165.31], [373.51, 165.81], [397.62, 165.79]]}]\n```" - } - ] - }, - { - "id": 318, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[488.99, 244.65], [520.82, 244.65], [516.82, 247.8], [483.41, 247.79], [488.99, 184.63], [520.82, 184.63], [516.82, 184.78], [483.41, 184.78]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[460.02, 245.04], [490.91, 245.04], [485.35, 248.25], [452.91, 248.25], [460.02, 185.99], [490.91, 185.99], [485.35, 186.21], [452.91, 186.21]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[497.07, 245.1], [470.78, 245.03], [476.55, 242.6], [501.85, 242.66], [497.07, 187.27], [470.78, 187.27], [476.55, 187.05], [501.85, 187.05]]}]\n```" - } - ] - }, - { - "id": 319, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[719.76, 252.73], [805.46, 252.48], [885.23, 277.52], [769.52, 277.98], [719.76, 159.69], [805.46, 159.77], [885.23, 152.1], [769.52, 151.96]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[375.62, 249.09], [405.66, 248.55], [403.82, 251.03], [372.65, 251.6], [375.62, 166.46], [405.66, 166.58], [403.82, 166.03], [372.65, 165.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[390.5, 246.95], [365.62, 246.89], [373.51, 244.87], [397.62, 244.93], [390.5, 165.29], [365.62, 165.31], [373.51, 165.81], [397.62, 165.79]]}]\n```" - } - ] - }, - { - "id": 320, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[719.76, 252.73], [805.46, 252.48], [885.23, 277.52], [769.52, 277.98], [719.76, 159.69], [805.46, 159.77], [885.23, 152.1], [769.52, 151.96]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[488.99, 244.65], [520.82, 244.65], [516.82, 247.8], [483.41, 247.79], [488.99, 184.63], [520.82, 184.63], [516.82, 184.78], [483.41, 184.78]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[460.02, 245.04], [490.91, 245.04], [485.35, 248.25], [452.91, 248.25], [460.02, 185.99], [490.91, 185.99], [485.35, 186.21], [452.91, 186.21]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[497.07, 245.1], [470.78, 245.03], [476.55, 242.6], [501.85, 242.66], [497.07, 187.27], [470.78, 187.27], [476.55, 187.05], [501.85, 187.05]]}]\n```" - } - ] - }, - { - "id": 321, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[488.99, 244.65], [520.82, 244.65], [516.82, 247.8], [483.41, 247.79], [488.99, 184.63], [520.82, 184.63], [516.82, 184.78], [483.41, 184.78]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[460.02, 245.04], [490.91, 245.04], [485.35, 248.25], [452.91, 248.25], [460.02, 185.99], [490.91, 185.99], [485.35, 186.21], [452.91, 186.21]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[497.07, 245.1], [470.78, 245.03], [476.55, 242.6], [501.85, 242.66], [497.07, 187.27], [470.78, 187.27], [476.55, 187.05], [501.85, 187.05]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[375.62, 249.09], [405.66, 248.55], [403.82, 251.03], [372.65, 251.6], [375.62, 166.46], [405.66, 166.58], [403.82, 166.03], [372.65, 165.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[390.5, 246.95], [365.62, 246.89], [373.51, 244.87], [397.62, 244.93], [390.5, 165.29], [365.62, 165.31], [373.51, 165.81], [397.62, 165.79]]}]\n```" - } - ] - }, - { - "id": 322, - "data_source": "Mono3DRefer", - "scene_name": "002045", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, pedestrian, person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[719.76, 252.73], [805.46, 252.48], [885.23, 277.52], [769.52, 277.98], [719.76, 159.69], [805.46, 159.77], [885.23, 152.1], [769.52, 151.96]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[488.99, 244.65], [520.82, 244.65], [516.82, 247.8], [483.41, 247.79], [488.99, 184.63], [520.82, 184.63], [516.82, 184.78], [483.41, 184.78]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[460.02, 245.04], [490.91, 245.04], [485.35, 248.25], [452.91, 248.25], [460.02, 185.99], [490.91, 185.99], [485.35, 186.21], [452.91, 186.21]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[497.07, 245.1], [470.78, 245.03], [476.55, 242.6], [501.85, 242.66], [497.07, 187.27], [470.78, 187.27], [476.55, 187.05], [501.85, 187.05]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[375.62, 249.09], [405.66, 248.55], [403.82, 251.03], [372.65, 251.6], [375.62, 166.46], [405.66, 166.58], [403.82, 166.03], [372.65, 165.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[390.5, 246.95], [365.62, 246.89], [373.51, 244.87], [397.62, 244.93], [390.5, 165.29], [365.62, 165.31], [373.51, 165.81], [397.62, 165.79]]}]\n```" - } - ] - }, - { - "id": 323, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[110.41, 405.12], [206.7, 406.38], [101.36, 462.37], [-18.02, 460.41], [110.41, 167.99], [206.7, 167.92], [101.36, 164.53], [-18.02, 164.65]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[922.82, 263.53], [972.86, 263.64], [1003.37, 270.65], [949.02, 270.51], [922.82, 155.09], [972.86, 155.06], [1003.37, 152.8], [949.02, 152.84]]}]\n```" - } - ] - }, - { - "id": 324, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"person_sitting\", \"corners_3d\": [[979.9, 378.24], [943.21, 358.8], [1034.07, 359.06], [1080.8, 378.56], [979.9, 150.08], [943.21, 153.19], [1034.07, 153.15], [1080.8, 150.03]]}]\n```" - } - ] - }, - { - "id": 325, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[683.24, 223.37], [733.6, 223.34], [758.8, 230.88], [699.37, 230.92], [683.24, 169.08], [733.6, 169.08], [758.8, 166.84], [699.37, 166.83]]}]\n```" - } - ] - }, - { - "id": 326, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, person_sitting). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[110.41, 405.12], [206.7, 406.38], [101.36, 462.37], [-18.02, 460.41], [110.41, 167.99], [206.7, 167.92], [101.36, 164.53], [-18.02, 164.65]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[922.82, 263.53], [972.86, 263.64], [1003.37, 270.65], [949.02, 270.51], [922.82, 155.09], [972.86, 155.06], [1003.37, 152.8], [949.02, 152.84]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[979.9, 378.24], [943.21, 358.8], [1034.07, 359.06], [1080.8, 378.56], [979.9, 150.08], [943.21, 153.19], [1034.07, 153.15], [1080.8, 150.03]]}]\n```" - } - ] - }, - { - "id": 327, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[110.41, 405.12], [206.7, 406.38], [101.36, 462.37], [-18.02, 460.41], [110.41, 167.99], [206.7, 167.92], [101.36, 164.53], [-18.02, 164.65]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[922.82, 263.53], [972.86, 263.64], [1003.37, 270.65], [949.02, 270.51], [922.82, 155.09], [972.86, 155.06], [1003.37, 152.8], [949.02, 152.84]]}, {\"category\": \"van\", \"corners_3d\": [[683.24, 223.37], [733.6, 223.34], [758.8, 230.88], [699.37, 230.92], [683.24, 169.08], [733.6, 169.08], [758.8, 166.84], [699.37, 166.83]]}]\n```" - } - ] - }, - { - "id": 328, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (person_sitting, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[683.24, 223.37], [733.6, 223.34], [758.8, 230.88], [699.37, 230.92], [683.24, 169.08], [733.6, 169.08], [758.8, 166.84], [699.37, 166.83]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[979.9, 378.24], [943.21, 358.8], [1034.07, 359.06], [1080.8, 378.56], [979.9, 150.08], [943.21, 153.19], [1034.07, 153.15], [1080.8, 150.03]]}]\n```" - } - ] - }, - { - "id": 329, - "data_source": "Mono3DRefer", - "scene_name": "002073", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, person_sitting, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[110.41, 405.12], [206.7, 406.38], [101.36, 462.37], [-18.02, 460.41], [110.41, 167.99], [206.7, 167.92], [101.36, 164.53], [-18.02, 164.65]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[922.82, 263.53], [972.86, 263.64], [1003.37, 270.65], [949.02, 270.51], [922.82, 155.09], [972.86, 155.06], [1003.37, 152.8], [949.02, 152.84]]}, {\"category\": \"van\", \"corners_3d\": [[683.24, 223.37], [733.6, 223.34], [758.8, 230.88], [699.37, 230.92], [683.24, 169.08], [733.6, 169.08], [758.8, 166.84], [699.37, 166.83]]}, {\"category\": \"person_sitting\", \"corners_3d\": [[979.9, 378.24], [943.21, 358.8], [1034.07, 359.06], [1080.8, 378.56], [979.9, 150.08], [943.21, 153.19], [1034.07, 153.15], [1080.8, 150.03]]}]\n```" - } - ] - }, - { - "id": 330, - "data_source": "Mono3DRefer", - "scene_name": "002103", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[824.06, 355.77], [984.74, 355.8], [1625.52, 668.74], [1189.81, 668.51], [824.06, 194.09], [984.74, 194.09], [1625.52, 230.42], [1189.81, 230.4]]}, {\"category\": \"car\", \"corners_3d\": [[469.5, 295.58], [370.13, 295.6], [433.31, 263.13], [506.39, 263.13], [469.5, 205.73], [370.13, 205.74], [433.31, 197.04], [506.39, 197.04]]}, {\"category\": \"car\", \"corners_3d\": [[700.92, 275.06], [798.45, 274.51], [911.81, 327.57], [763.89, 328.84], [700.92, 189.07], [798.45, 188.98], [911.81, 197.4], [763.89, 197.6]]}, {\"category\": \"car\", \"corners_3d\": [[556.56, 216.6], [522.81, 216.66], [529.17, 212.35], [559.59, 212.3], [556.56, 186.26], [522.81, 186.28], [529.17, 184.96], [559.59, 184.94]]}, {\"category\": \"car\", \"corners_3d\": [[635.63, 205.47], [663.79, 205.46], [670.26, 208.92], [639.1, 208.94], [635.63, 180.02], [663.79, 180.02], [670.26, 180.78], [639.1, 180.78]]}]\n```" - } - ] - }, - { - "id": 331, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1378.1, 414.72], [1111.48, 421.58], [833.44, 296.54], [971.75, 294.82], [1378.1, 122.12], [1111.48, 120.68], [833.44, 146.91], [971.75, 147.27]]}, {\"category\": \"car\", \"corners_3d\": [[218.74, 406.07], [25.23, 407.96], [260.64, 309.5], [372.38, 308.86], [218.74, 222.25], [25.23, 222.65], [260.64, 201.8], [372.38, 201.66]]}, {\"category\": \"car\", \"corners_3d\": [[453.01, 268.97], [373.7, 269.4], [415.41, 249.89], [478.61, 249.61], [453.01, 204.37], [373.7, 204.51], [415.41, 198.11], [478.61, 198.02]]}, {\"category\": \"car\", \"corners_3d\": [[494.77, 240.82], [435.64, 240.99], [460.21, 229.95], [509.73, 229.83], [494.77, 182.81], [435.64, 182.83], [460.21, 181.22], [509.73, 181.2]]}, {\"category\": \"car\", \"corners_3d\": [[601.83, 206.92], [629.78, 206.86], [634.64, 209.82], [604.26, 209.9], [601.83, 180.17], [629.78, 180.16], [634.64, 180.8], [604.26, 180.81]]}]\n```" - } - ] - }, - { - "id": 332, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[564.27, 205.95], [532.12, 205.88], [546.56, 201.66], [574.62, 201.72], [564.27, 157.05], [532.12, 157.08], [546.56, 159.1], [574.62, 159.07]]}]\n```" - } - ] - }, - { - "id": 333, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[594.82, 208.73], [602.26, 208.43], [607.21, 208.74], [599.74, 209.04], [594.82, 176.83], [602.26, 176.8], [607.21, 176.84], [599.74, 176.87]]}]\n```" - } - ] - }, - { - "id": 334, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1378.1, 414.72], [1111.48, 421.58], [833.44, 296.54], [971.75, 294.82], [1378.1, 122.12], [1111.48, 120.68], [833.44, 146.91], [971.75, 147.27]]}, {\"category\": \"car\", \"corners_3d\": [[218.74, 406.07], [25.23, 407.96], [260.64, 309.5], [372.38, 308.86], [218.74, 222.25], [25.23, 222.65], [260.64, 201.8], [372.38, 201.66]]}, {\"category\": \"car\", \"corners_3d\": [[453.01, 268.97], [373.7, 269.4], [415.41, 249.89], [478.61, 249.61], [453.01, 204.37], [373.7, 204.51], [415.41, 198.11], [478.61, 198.02]]}, {\"category\": \"car\", \"corners_3d\": [[494.77, 240.82], [435.64, 240.99], [460.21, 229.95], [509.73, 229.83], [494.77, 182.81], [435.64, 182.83], [460.21, 181.22], [509.73, 181.2]]}, {\"category\": \"truck\", \"corners_3d\": [[564.27, 205.95], [532.12, 205.88], [546.56, 201.66], [574.62, 201.72], [564.27, 157.05], [532.12, 157.08], [546.56, 159.1], [574.62, 159.07]]}, {\"category\": \"car\", \"corners_3d\": [[601.83, 206.92], [629.78, 206.86], [634.64, 209.82], [604.26, 209.9], [601.83, 180.17], [629.78, 180.16], [634.64, 180.8], [604.26, 180.81]]}]\n```" - } - ] - }, - { - "id": 335, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1378.1, 414.72], [1111.48, 421.58], [833.44, 296.54], [971.75, 294.82], [1378.1, 122.12], [1111.48, 120.68], [833.44, 146.91], [971.75, 147.27]]}, {\"category\": \"car\", \"corners_3d\": [[218.74, 406.07], [25.23, 407.96], [260.64, 309.5], [372.38, 308.86], [218.74, 222.25], [25.23, 222.65], [260.64, 201.8], [372.38, 201.66]]}, {\"category\": \"car\", \"corners_3d\": [[453.01, 268.97], [373.7, 269.4], [415.41, 249.89], [478.61, 249.61], [453.01, 204.37], [373.7, 204.51], [415.41, 198.11], [478.61, 198.02]]}, {\"category\": \"car\", \"corners_3d\": [[494.77, 240.82], [435.64, 240.99], [460.21, 229.95], [509.73, 229.83], [494.77, 182.81], [435.64, 182.83], [460.21, 181.22], [509.73, 181.2]]}, {\"category\": \"car\", \"corners_3d\": [[601.83, 206.92], [629.78, 206.86], [634.64, 209.82], [604.26, 209.9], [601.83, 180.17], [629.78, 180.16], [634.64, 180.8], [604.26, 180.81]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[594.82, 208.73], [602.26, 208.43], [607.21, 208.74], [599.74, 209.04], [594.82, 176.83], [602.26, 176.8], [607.21, 176.84], [599.74, 176.87]]}]\n```" - } - ] - }, - { - "id": 336, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[564.27, 205.95], [532.12, 205.88], [546.56, 201.66], [574.62, 201.72], [564.27, 157.05], [532.12, 157.08], [546.56, 159.1], [574.62, 159.07]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[594.82, 208.73], [602.26, 208.43], [607.21, 208.74], [599.74, 209.04], [594.82, 176.83], [602.26, 176.8], [607.21, 176.84], [599.74, 176.87]]}]\n```" - } - ] - }, - { - "id": 337, - "data_source": "Mono3DRefer", - "scene_name": "002128", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1378.1, 414.72], [1111.48, 421.58], [833.44, 296.54], [971.75, 294.82], [1378.1, 122.12], [1111.48, 120.68], [833.44, 146.91], [971.75, 147.27]]}, {\"category\": \"car\", \"corners_3d\": [[218.74, 406.07], [25.23, 407.96], [260.64, 309.5], [372.38, 308.86], [218.74, 222.25], [25.23, 222.65], [260.64, 201.8], [372.38, 201.66]]}, {\"category\": \"car\", \"corners_3d\": [[453.01, 268.97], [373.7, 269.4], [415.41, 249.89], [478.61, 249.61], [453.01, 204.37], [373.7, 204.51], [415.41, 198.11], [478.61, 198.02]]}, {\"category\": \"car\", \"corners_3d\": [[494.77, 240.82], [435.64, 240.99], [460.21, 229.95], [509.73, 229.83], [494.77, 182.81], [435.64, 182.83], [460.21, 181.22], [509.73, 181.2]]}, {\"category\": \"truck\", \"corners_3d\": [[564.27, 205.95], [532.12, 205.88], [546.56, 201.66], [574.62, 201.72], [564.27, 157.05], [532.12, 157.08], [546.56, 159.1], [574.62, 159.07]]}, {\"category\": \"car\", \"corners_3d\": [[601.83, 206.92], [629.78, 206.86], [634.64, 209.82], [604.26, 209.9], [601.83, 180.17], [629.78, 180.16], [634.64, 180.8], [604.26, 180.81]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[594.82, 208.73], [602.26, 208.43], [607.21, 208.74], [599.74, 209.04], [594.82, 176.83], [602.26, 176.8], [607.21, 176.84], [599.74, 176.87]]}]\n```" - } - ] - }, - { - "id": 338, - "data_source": "Mono3DRefer", - "scene_name": "002188", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[411.29, 209.66], [382.36, 209.68], [401.87, 206.4], [428.22, 206.39], [411.29, 185.89], [382.36, 185.89], [401.87, 184.73], [428.22, 184.73]]}]\n```" - } - ] - }, - { - "id": 339, - "data_source": "Mono3DRefer", - "scene_name": "002188", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[282.26, 218.24], [230.12, 218.27], [295.76, 210.25], [338.66, 210.23], [282.26, 145.96], [230.12, 145.94], [295.76, 150.69], [338.66, 150.71]]}, {\"category\": \"truck\", \"corners_3d\": [[345.62, 209.66], [304.45, 209.66], [348.3, 204.36], [383.54, 204.36], [345.62, 151.01], [304.45, 151.01], [348.3, 154.16], [383.54, 154.16]]}]\n```" - } - ] - }, - { - "id": 340, - "data_source": "Mono3DRefer", - "scene_name": "002188", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[282.26, 218.24], [230.12, 218.27], [295.76, 210.25], [338.66, 210.23], [282.26, 145.96], [230.12, 145.94], [295.76, 150.69], [338.66, 150.71]]}, {\"category\": \"truck\", \"corners_3d\": [[345.62, 209.66], [304.45, 209.66], [348.3, 204.36], [383.54, 204.36], [345.62, 151.01], [304.45, 151.01], [348.3, 154.16], [383.54, 154.16]]}, {\"category\": \"car\", \"corners_3d\": [[411.29, 209.66], [382.36, 209.68], [401.87, 206.4], [428.22, 206.39], [411.29, 185.89], [382.36, 185.89], [401.87, 184.73], [428.22, 184.73]]}]\n```" - } - ] - }, - { - "id": 341, - "data_source": "Mono3DRefer", - "scene_name": "002250", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[398.11, 304.34], [271.58, 305.72], [361.75, 264.98], [449.11, 264.31], [398.11, 195.37], [271.58, 195.61], [361.75, 188.63], [449.11, 188.52]]}, {\"category\": \"car\", \"corners_3d\": [[710.6, 235.94], [655.23, 236.24], [640.89, 226.09], [687.49, 225.88], [710.6, 178.83], [655.23, 178.86], [640.89, 177.9], [687.49, 177.88]]}, {\"category\": \"car\", \"corners_3d\": [[520.08, 223.07], [474.97, 223.08], [494.03, 215.93], [532.72, 215.93], [520.08, 183.92], [474.97, 183.92], [494.03, 182.34], [532.72, 182.34]]}, {\"category\": \"car\", \"corners_3d\": [[624.55, 217.88], [661.13, 217.62], [678.23, 223.48], [636.94, 223.8], [624.55, 183.58], [661.13, 183.52], [678.23, 184.91], [636.94, 184.99]]}]\n```" - } - ] - }, - { - "id": 342, - "data_source": "Mono3DRefer", - "scene_name": "002300", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-884.1, 1500.51], [-1962.83, 1460.81], [112.9, 427.09], [333.02, 428.6], [-884.1, 452.93], [-1962.83, 444.55], [112.9, 226.49], [333.02, 226.81]]}, {\"category\": \"car\", \"corners_3d\": [[549.36, 232.14], [499.9, 232.23], [512.96, 224.03], [555.59, 223.97], [549.36, 187.67], [499.9, 187.69], [512.96, 185.64], [555.59, 185.63]]}, {\"category\": \"car\", \"corners_3d\": [[659.15, 216.39], [696.58, 216.39], [709.22, 222.76], [666.32, 222.76], [659.15, 182.6], [696.58, 182.6], [709.22, 184.02], [666.32, 184.02]]}, {\"category\": \"car\", \"corners_3d\": [[555.4, 207.51], [580.24, 207.5], [577.99, 210.95], [550.68, 210.96], [555.4, 183.65], [580.24, 183.65], [577.99, 184.72], [550.68, 184.73]]}]\n```" - } - ] - }, - { - "id": 343, - "data_source": "Mono3DRefer", - "scene_name": "002307", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[741.7, 212.54], [726.25, 212.56], [723.87, 211.84], [739.05, 211.83], [741.7, 176.57], [726.25, 176.57], [723.87, 176.5], [739.05, 176.5]]}]\n```" - } - ] - }, - { - "id": 344, - "data_source": "Mono3DRefer", - "scene_name": "002307", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[102.8, 255.61], [32.17, 255.77], [129.48, 241.43], [187.75, 241.31], [102.8, 193.14], [32.17, 193.18], [129.48, 189.66], [187.75, 189.64]]}, {\"category\": \"car\", \"corners_3d\": [[69.86, 245.85], [9.71, 245.85], [101.64, 234.66], [152.56, 234.65], [69.86, 189.82], [9.71, 189.82], [101.64, 187.22], [152.56, 187.22]]}, {\"category\": \"car\", \"corners_3d\": [[375.14, 211.4], [343.45, 211.33], [368.89, 208.0], [397.88, 208.06], [375.14, 183.46], [343.45, 183.44], [368.89, 182.52], [397.88, 182.54]]}]\n```" - } - ] - }, - { - "id": 345, - "data_source": "Mono3DRefer", - "scene_name": "002307", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[102.8, 255.61], [32.17, 255.77], [129.48, 241.43], [187.75, 241.31], [102.8, 193.14], [32.17, 193.18], [129.48, 189.66], [187.75, 189.64]]}, {\"category\": \"car\", \"corners_3d\": [[69.86, 245.85], [9.71, 245.85], [101.64, 234.66], [152.56, 234.65], [69.86, 189.82], [9.71, 189.82], [101.64, 187.22], [152.56, 187.22]]}, {\"category\": \"car\", \"corners_3d\": [[375.14, 211.4], [343.45, 211.33], [368.89, 208.0], [397.88, 208.06], [375.14, 183.46], [343.45, 183.44], [368.89, 182.52], [397.88, 182.54]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[741.7, 212.54], [726.25, 212.56], [723.87, 211.84], [739.05, 211.83], [741.7, 176.57], [726.25, 176.57], [723.87, 176.5], [739.05, 176.5]]}]\n```" - } - ] - }, - { - "id": 346, - "data_source": "Mono3DRefer", - "scene_name": "002334", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[444.02, 292.98], [433.08, 300.66], [343.07, 300.66], [359.75, 292.98], [444.02, 158.16], [433.08, 156.63], [343.07, 156.63], [359.75, 158.16]]}]\n```" - } - ] - }, - { - "id": 347, - "data_source": "Mono3DRefer", - "scene_name": "002376", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[323.25, 358.42], [141.88, 357.99], [327.76, 285.44], [438.22, 285.6], [323.25, 209.31], [141.88, 209.23], [327.76, 194.97], [438.22, 195.0]]}, {\"category\": \"car\", \"corners_3d\": [[547.35, 207.59], [514.52, 207.62], [522.8, 204.02], [552.23, 204.0], [547.35, 176.83], [514.52, 176.83], [522.8, 176.42], [552.23, 176.42]]}]\n```" - } - ] - }, - { - "id": 348, - "data_source": "Mono3DRefer", - "scene_name": "002376", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[460.68, 255.15], [383.06, 254.98], [431.87, 238.31], [493.78, 238.41], [460.68, 174.24], [383.06, 174.24], [431.87, 173.96], [493.78, 173.96]]}, {\"category\": \"van\", \"corners_3d\": [[556.28, 198.76], [525.33, 198.76], [533.25, 196.3], [561.27, 196.3], [556.28, 165.02], [525.33, 165.02], [533.25, 165.76], [561.27, 165.76]]}]\n```" - } - ] - }, - { - "id": 349, - "data_source": "Mono3DRefer", - "scene_name": "002376", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[323.25, 358.42], [141.88, 357.99], [327.76, 285.44], [438.22, 285.6], [323.25, 209.31], [141.88, 209.23], [327.76, 194.97], [438.22, 195.0]]}, {\"category\": \"van\", \"corners_3d\": [[460.68, 255.15], [383.06, 254.98], [431.87, 238.31], [493.78, 238.41], [460.68, 174.24], [383.06, 174.24], [431.87, 173.96], [493.78, 173.96]]}, {\"category\": \"car\", \"corners_3d\": [[547.35, 207.59], [514.52, 207.62], [522.8, 204.02], [552.23, 204.0], [547.35, 176.83], [514.52, 176.83], [522.8, 176.42], [552.23, 176.42]]}, {\"category\": \"van\", \"corners_3d\": [[556.28, 198.76], [525.33, 198.76], [533.25, 196.3], [561.27, 196.3], [556.28, 165.02], [525.33, 165.02], [533.25, 165.76], [561.27, 165.76]]}]\n```" - } - ] - }, - { - "id": 350, - "data_source": "Mono3DRefer", - "scene_name": "002383", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-2187.79, 2035.39], [-4314.71, 2141.06], [-103.01, 452.52], [180.12, 450.28], [-2187.79, 422.35], [-4314.71, 436.51], [-103.01, 210.32], [180.12, 210.02]]}, {\"category\": \"car\", \"corners_3d\": [[839.63, 273.01], [731.73, 273.32], [695.74, 246.96], [775.48, 246.79], [839.63, 176.89], [731.73, 176.9], [695.74, 175.84], [775.48, 175.83]]}, {\"category\": \"car\", \"corners_3d\": [[649.02, 218.76], [614.67, 216.09], [708.2, 211.79], [744.13, 213.94], [649.02, 165.36], [614.67, 165.8], [708.2, 166.5], [744.13, 166.15]]}]\n```" - } - ] - }, - { - "id": 351, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[432.48, 199.65], [411.03, 199.65], [423.59, 197.95], [443.67, 197.95], [432.48, 181.5], [411.03, 181.5], [423.59, 180.95], [443.67, 180.95]]}]\n```" - } - ] - }, - { - "id": 352, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[75.22, 290.32], [-103.87, 290.64], [331.9, 217.9], [400.06, 217.85], [75.22, 76.52], [-103.87, 76.27], [331.9, 135.92], [400.06, 135.95]]}]\n```" - } - ] - }, - { - "id": 353, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[875.43, 239.22], [898.53, 239.2], [929.83, 246.22], [904.29, 246.24], [875.43, 159.65], [898.53, 159.66], [929.83, 158.26], [904.29, 158.25]]}]\n```" - } - ] - }, - { - "id": 354, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[75.22, 290.32], [-103.87, 290.64], [331.9, 217.9], [400.06, 217.85], [75.22, 76.52], [-103.87, 76.27], [331.9, 135.92], [400.06, 135.95]]}, {\"category\": \"car\", \"corners_3d\": [[432.48, 199.65], [411.03, 199.65], [423.59, 197.95], [443.67, 197.95], [432.48, 181.5], [411.03, 181.5], [423.59, 180.95], [443.67, 180.95]]}]\n```" - } - ] - }, - { - "id": 355, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[875.43, 239.22], [898.53, 239.2], [929.83, 246.22], [904.29, 246.24], [875.43, 159.65], [898.53, 159.66], [929.83, 158.26], [904.29, 158.25]]}, {\"category\": \"car\", \"corners_3d\": [[432.48, 199.65], [411.03, 199.65], [423.59, 197.95], [443.67, 197.95], [432.48, 181.5], [411.03, 181.5], [423.59, 180.95], [443.67, 180.95]]}]\n```" - } - ] - }, - { - "id": 356, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[75.22, 290.32], [-103.87, 290.64], [331.9, 217.9], [400.06, 217.85], [75.22, 76.52], [-103.87, 76.27], [331.9, 135.92], [400.06, 135.95]]}, {\"category\": \"cyclist\", \"corners_3d\": [[875.43, 239.22], [898.53, 239.2], [929.83, 246.22], [904.29, 246.24], [875.43, 159.65], [898.53, 159.66], [929.83, 158.26], [904.29, 158.25]]}]\n```" - } - ] - }, - { - "id": 357, - "data_source": "Mono3DRefer", - "scene_name": "002391", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[75.22, 290.32], [-103.87, 290.64], [331.9, 217.9], [400.06, 217.85], [75.22, 76.52], [-103.87, 76.27], [331.9, 135.92], [400.06, 135.95]]}, {\"category\": \"cyclist\", \"corners_3d\": [[875.43, 239.22], [898.53, 239.2], [929.83, 246.22], [904.29, 246.24], [875.43, 159.65], [898.53, 159.66], [929.83, 158.26], [904.29, 158.25]]}, {\"category\": \"car\", \"corners_3d\": [[432.48, 199.65], [411.03, 199.65], [423.59, 197.95], [443.67, 197.95], [432.48, 181.5], [411.03, 181.5], [423.59, 180.95], [443.67, 180.95]]}]\n```" - } - ] - }, - { - "id": 358, - "data_source": "Mono3DRefer", - "scene_name": "002439", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.41, 282.81], [866.54, 282.05], [1039.26, 346.53], [881.28, 348.46], [766.41, 177.72], [866.54, 177.69], [1039.26, 180.54], [881.28, 180.63]]}, {\"category\": \"car\", \"corners_3d\": [[526.16, 283.94], [436.34, 283.95], [476.83, 257.9], [545.59, 257.89], [526.16, 201.04], [436.34, 201.05], [476.83, 194.43], [545.59, 194.43]]}, {\"category\": \"car\", \"corners_3d\": [[544.21, 218.59], [577.69, 218.47], [579.25, 225.0], [540.97, 225.17], [544.21, 186.22], [577.69, 186.19], [579.25, 188.1], [540.97, 188.15]]}, {\"category\": \"car\", \"corners_3d\": [[656.97, 220.44], [692.69, 220.37], [705.61, 226.26], [665.48, 226.35], [656.97, 180.89], [692.69, 180.88], [705.61, 181.87], [665.48, 181.88]]}]\n```" - } - ] - }, - { - "id": 359, - "data_source": "Mono3DRefer", - "scene_name": "002463", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[460.29, 211.17], [426.54, 211.36], [436.56, 207.61], [466.99, 207.45], [460.29, 181.61], [426.54, 181.66], [436.56, 180.8], [466.99, 180.76]]}]\n```" - } - ] - }, - { - "id": 360, - "data_source": "Mono3DRefer", - "scene_name": "002483", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[101.5, 238.52], [19.52, 238.6], [166.08, 222.05], [227.31, 222.0], [101.5, 125.23], [19.52, 125.17], [166.08, 137.18], [227.31, 137.21]]}, {\"category\": \"truck\", \"corners_3d\": [[240.05, 220.9], [181.88, 220.9], [263.89, 211.68], [310.88, 211.67], [240.05, 138.04], [181.88, 138.04], [263.89, 144.73], [310.88, 144.73]]}, {\"category\": \"truck\", \"corners_3d\": [[420.84, 198.18], [394.76, 198.18], [415.05, 195.78], [438.65, 195.78], [420.84, 163.35], [394.76, 163.35], [415.05, 164.25], [438.65, 164.25]]}]\n```" - } - ] - }, - { - "id": 361, - "data_source": "Mono3DRefer", - "scene_name": "002483", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[301.84, 223.21], [256.3, 223.21], [303.38, 216.49], [342.84, 216.49], [301.84, 185.68], [256.3, 185.68], [303.38, 183.97], [342.84, 183.97]]}]\n```" - } - ] - }, - { - "id": 362, - "data_source": "Mono3DRefer", - "scene_name": "002483", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[301.84, 223.21], [256.3, 223.21], [303.38, 216.49], [342.84, 216.49], [301.84, 185.68], [256.3, 185.68], [303.38, 183.97], [342.84, 183.97]]}, {\"category\": \"truck\", \"corners_3d\": [[101.5, 238.52], [19.52, 238.6], [166.08, 222.05], [227.31, 222.0], [101.5, 125.23], [19.52, 125.17], [166.08, 137.18], [227.31, 137.21]]}, {\"category\": \"truck\", \"corners_3d\": [[240.05, 220.9], [181.88, 220.9], [263.89, 211.68], [310.88, 211.67], [240.05, 138.04], [181.88, 138.04], [263.89, 144.73], [310.88, 144.73]]}, {\"category\": \"truck\", \"corners_3d\": [[420.84, 198.18], [394.76, 198.18], [415.05, 195.78], [438.65, 195.78], [420.84, 163.35], [394.76, 163.35], [415.05, 164.25], [438.65, 164.25]]}]\n```" - } - ] - }, - { - "id": 363, - "data_source": "Mono3DRefer", - "scene_name": "002504", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[650.22, 206.28], [673.91, 206.29], [680.26, 209.62], [654.21, 209.61], [650.22, 185.14], [673.91, 185.14], [680.26, 186.37], [654.21, 186.37]]}]\n```" - } - ] - }, - { - "id": 364, - "data_source": "Mono3DRefer", - "scene_name": "002539", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[567.95, 190.77], [585.61, 190.53], [610.52, 191.73], [591.99, 191.99], [567.95, 164.28], [585.61, 164.39], [610.52, 163.82], [591.99, 163.7]]}]\n```" - } - ] - }, - { - "id": 365, - "data_source": "Mono3DRefer", - "scene_name": "002539", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[439.45, 287.92], [345.57, 287.63], [410.69, 260.73], [482.65, 260.9], [439.45, 201.45], [345.57, 201.38], [410.69, 194.69], [482.65, 194.73]]}, {\"category\": \"car\", \"corners_3d\": [[548.78, 209.67], [520.68, 209.75], [525.77, 206.15], [551.12, 206.09], [548.78, 186.83], [520.68, 186.86], [525.77, 185.49], [551.12, 185.47]]}, {\"category\": \"car\", \"corners_3d\": [[409.85, 216.27], [406.3, 214.7], [489.77, 212.91], [496.16, 214.35], [409.85, 184.69], [406.3, 184.26], [489.77, 183.77], [496.16, 184.17]]}]\n```" - } - ] - }, - { - "id": 366, - "data_source": "Mono3DRefer", - "scene_name": "002539", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[439.45, 287.92], [345.57, 287.63], [410.69, 260.73], [482.65, 260.9], [439.45, 201.45], [345.57, 201.38], [410.69, 194.69], [482.65, 194.73]]}, {\"category\": \"car\", \"corners_3d\": [[548.78, 209.67], [520.68, 209.75], [525.77, 206.15], [551.12, 206.09], [548.78, 186.83], [520.68, 186.86], [525.77, 185.49], [551.12, 185.47]]}, {\"category\": \"van\", \"corners_3d\": [[567.95, 190.77], [585.61, 190.53], [610.52, 191.73], [591.99, 191.99], [567.95, 164.28], [585.61, 164.39], [610.52, 163.82], [591.99, 163.7]]}, {\"category\": \"car\", \"corners_3d\": [[409.85, 216.27], [406.3, 214.7], [489.77, 212.91], [496.16, 214.35], [409.85, 184.69], [406.3, 184.26], [489.77, 183.77], [496.16, 184.17]]}]\n```" - } - ] - }, - { - "id": 367, - "data_source": "Mono3DRefer", - "scene_name": "002570", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[906.84, 309.08], [775.4, 309.6], [716.4, 264.99], [805.23, 264.75], [906.84, 191.1], [775.4, 191.17], [716.4, 185.2], [805.23, 185.16]]}, {\"category\": \"car\", \"corners_3d\": [[427.22, 299.96], [304.25, 300.41], [383.46, 265.6], [472.76, 265.36], [427.22, 194.39], [304.25, 194.46], [383.46, 188.57], [472.76, 188.53]]}, {\"category\": \"car\", \"corners_3d\": [[680.57, 235.6], [741.67, 235.55], [775.57, 250.87], [699.56, 250.94], [680.57, 182.14], [741.67, 182.13], [775.57, 184.4], [699.56, 184.41]]}, {\"category\": \"car\", \"corners_3d\": [[734.31, 229.69], [675.63, 229.69], [664.89, 220.53], [714.11, 220.53], [734.31, 177.61], [675.63, 177.61], [664.89, 176.84], [714.11, 176.84]]}, {\"category\": \"car\", \"corners_3d\": [[747.39, 201.64], [739.2, 200.29], [821.78, 200.13], [833.95, 201.46], [747.39, 167.9], [739.2, 168.14], [821.78, 168.16], [833.95, 167.94]]}]\n```" - } - ] - }, - { - "id": 368, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[577.41, 196.59], [553.23, 196.53], [561.08, 194.89], [583.6, 194.95], [577.41, 172.05], [553.23, 172.06], [561.08, 172.11], [583.6, 172.11]]}]\n```" - } - ] - }, - { - "id": 369, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[429.65, 429.9], [348.19, 429.63], [400.9, 379.13], [466.37, 379.3], [429.65, 160.91], [348.19, 160.92], [400.9, 163.27], [466.37, 163.26]]}]\n```" - } - ] - }, - { - "id": 370, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[464.54, 286.17], [357.72, 286.18], [425.65, 255.55], [503.59, 255.54], [464.54, 174.8], [357.72, 174.8], [425.65, 174.27], [503.59, 174.27]]}, {\"category\": \"car\", \"corners_3d\": [[779.83, 339.64], [939.87, 340.77], [1187.81, 475.91], [897.54, 472.24], [779.83, 203.6], [939.87, 203.81], [1187.81, 228.72], [897.54, 228.05]]}, {\"category\": \"car\", \"corners_3d\": [[493.94, 227.7], [546.42, 227.82], [529.82, 238.51], [467.2, 238.33], [493.94, 183.56], [546.42, 183.59], [529.82, 185.67], [467.2, 185.64]]}, {\"category\": \"car\", \"corners_3d\": [[668.3, 209.18], [635.98, 209.18], [633.44, 205.76], [662.71, 205.76], [668.3, 179.27], [635.98, 179.27], [633.44, 178.67], [662.71, 178.67]]}, {\"category\": \"car\", \"corners_3d\": [[-536.61, 369.63], [-290.39, 327.05], [137.18, 326.01], [10.14, 367.94], [-536.61, 175.26], [-290.39, 174.74], [137.18, 174.72], [10.14, 175.23]]}]\n```" - } - ] - }, - { - "id": 371, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[429.65, 429.9], [348.19, 429.63], [400.9, 379.13], [466.37, 379.3], [429.65, 160.91], [348.19, 160.92], [400.9, 163.27], [466.37, 163.26]]}, {\"category\": \"van\", \"corners_3d\": [[577.41, 196.59], [553.23, 196.53], [561.08, 194.89], [583.6, 194.95], [577.41, 172.05], [553.23, 172.06], [561.08, 172.11], [583.6, 172.11]]}]\n```" - } - ] - }, - { - "id": 372, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[464.54, 286.17], [357.72, 286.18], [425.65, 255.55], [503.59, 255.54], [464.54, 174.8], [357.72, 174.8], [425.65, 174.27], [503.59, 174.27]]}, {\"category\": \"car\", \"corners_3d\": [[779.83, 339.64], [939.87, 340.77], [1187.81, 475.91], [897.54, 472.24], [779.83, 203.6], [939.87, 203.81], [1187.81, 228.72], [897.54, 228.05]]}, {\"category\": \"car\", \"corners_3d\": [[493.94, 227.7], [546.42, 227.82], [529.82, 238.51], [467.2, 238.33], [493.94, 183.56], [546.42, 183.59], [529.82, 185.67], [467.2, 185.64]]}, {\"category\": \"car\", \"corners_3d\": [[668.3, 209.18], [635.98, 209.18], [633.44, 205.76], [662.71, 205.76], [668.3, 179.27], [635.98, 179.27], [633.44, 178.67], [662.71, 178.67]]}, {\"category\": \"van\", \"corners_3d\": [[577.41, 196.59], [553.23, 196.53], [561.08, 194.89], [583.6, 194.95], [577.41, 172.05], [553.23, 172.06], [561.08, 172.11], [583.6, 172.11]]}, {\"category\": \"car\", \"corners_3d\": [[-536.61, 369.63], [-290.39, 327.05], [137.18, 326.01], [10.14, 367.94], [-536.61, 175.26], [-290.39, 174.74], [137.18, 174.72], [10.14, 175.23]]}]\n```" - } - ] - }, - { - "id": 373, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[464.54, 286.17], [357.72, 286.18], [425.65, 255.55], [503.59, 255.54], [464.54, 174.8], [357.72, 174.8], [425.65, 174.27], [503.59, 174.27]]}, {\"category\": \"cyclist\", \"corners_3d\": [[429.65, 429.9], [348.19, 429.63], [400.9, 379.13], [466.37, 379.3], [429.65, 160.91], [348.19, 160.92], [400.9, 163.27], [466.37, 163.26]]}, {\"category\": \"car\", \"corners_3d\": [[779.83, 339.64], [939.87, 340.77], [1187.81, 475.91], [897.54, 472.24], [779.83, 203.6], [939.87, 203.81], [1187.81, 228.72], [897.54, 228.05]]}, {\"category\": \"car\", \"corners_3d\": [[493.94, 227.7], [546.42, 227.82], [529.82, 238.51], [467.2, 238.33], [493.94, 183.56], [546.42, 183.59], [529.82, 185.67], [467.2, 185.64]]}, {\"category\": \"car\", \"corners_3d\": [[668.3, 209.18], [635.98, 209.18], [633.44, 205.76], [662.71, 205.76], [668.3, 179.27], [635.98, 179.27], [633.44, 178.67], [662.71, 178.67]]}, {\"category\": \"car\", \"corners_3d\": [[-536.61, 369.63], [-290.39, 327.05], [137.18, 326.01], [10.14, 367.94], [-536.61, 175.26], [-290.39, 174.74], [137.18, 174.72], [10.14, 175.23]]}]\n```" - } - ] - }, - { - "id": 374, - "data_source": "Mono3DRefer", - "scene_name": "002612", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[464.54, 286.17], [357.72, 286.18], [425.65, 255.55], [503.59, 255.54], [464.54, 174.8], [357.72, 174.8], [425.65, 174.27], [503.59, 174.27]]}, {\"category\": \"cyclist\", \"corners_3d\": [[429.65, 429.9], [348.19, 429.63], [400.9, 379.13], [466.37, 379.3], [429.65, 160.91], [348.19, 160.92], [400.9, 163.27], [466.37, 163.26]]}, {\"category\": \"car\", \"corners_3d\": [[779.83, 339.64], [939.87, 340.77], [1187.81, 475.91], [897.54, 472.24], [779.83, 203.6], [939.87, 203.81], [1187.81, 228.72], [897.54, 228.05]]}, {\"category\": \"car\", \"corners_3d\": [[493.94, 227.7], [546.42, 227.82], [529.82, 238.51], [467.2, 238.33], [493.94, 183.56], [546.42, 183.59], [529.82, 185.67], [467.2, 185.64]]}, {\"category\": \"car\", \"corners_3d\": [[668.3, 209.18], [635.98, 209.18], [633.44, 205.76], [662.71, 205.76], [668.3, 179.27], [635.98, 179.27], [633.44, 178.67], [662.71, 178.67]]}, {\"category\": \"van\", \"corners_3d\": [[577.41, 196.59], [553.23, 196.53], [561.08, 194.89], [583.6, 194.95], [577.41, 172.05], [553.23, 172.06], [561.08, 172.11], [583.6, 172.11]]}, {\"category\": \"car\", \"corners_3d\": [[-536.61, 369.63], [-290.39, 327.05], [137.18, 326.01], [10.14, 367.94], [-536.61, 175.26], [-290.39, 174.74], [137.18, 174.72], [10.14, 175.23]]}]\n```" - } - ] - }, - { - "id": 375, - "data_source": "Mono3DRefer", - "scene_name": "002695", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[656.58, 211.77], [682.64, 211.77], [690.67, 216.08], [661.73, 216.07], [656.58, 188.51], [682.64, 188.52], [690.67, 190.25], [661.73, 190.25]]}]\n```" - } - ] - }, - { - "id": 376, - "data_source": "Mono3DRefer", - "scene_name": "002729", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[2985.01, 1240.09], [1959.3, 1273.96], [866.33, 392.13], [1082.62, 390.75], [2985.01, 286.75], [1959.3, 290.36], [866.33, 196.25], [1082.62, 196.11]]}, {\"category\": \"car\", \"corners_3d\": [[-281.01, 773.04], [-899.76, 783.48], [70.34, 387.06], [283.96, 385.76], [-281.01, 259.52], [-899.76, 261.02], [70.34, 203.78], [283.96, 203.6]]}, {\"category\": \"car\", \"corners_3d\": [[728.41, 276.9], [832.84, 276.76], [949.06, 329.28], [791.97, 329.6], [728.41, 185.42], [832.84, 185.41], [949.06, 191.75], [791.97, 191.79]]}, {\"category\": \"car\", \"corners_3d\": [[811.82, 264.85], [714.63, 265.11], [685.52, 242.74], [759.25, 242.59], [811.82, 178.32], [714.63, 178.33], [685.52, 177.0], [759.25, 177.0]]}, {\"category\": \"car\", \"corners_3d\": [[664.67, 213.94], [634.94, 211.63], [725.01, 208.37], [756.99, 210.3], [664.67, 165.06], [634.94, 165.5], [725.01, 166.12], [756.99, 165.75]]}]\n```" - } - ] - }, - { - "id": 377, - "data_source": "Mono3DRefer", - "scene_name": "002801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[172.29, 241.15], [207.74, 235.6], [332.96, 235.58], [308.6, 241.13], [172.29, 175.69], [207.74, 175.46], [332.96, 175.45], [308.6, 175.68]]}, {\"category\": \"car\", \"corners_3d\": [[738.66, 203.68], [698.49, 203.83], [679.13, 199.96], [714.38, 199.83], [738.66, 161.22], [698.49, 161.16], [679.13, 162.63], [714.38, 162.67]]}, {\"category\": \"car\", \"corners_3d\": [[552.57, 210.1], [518.29, 210.23], [522.88, 205.96], [553.23, 205.86], [552.57, 176.96], [518.29, 176.97], [522.88, 176.5], [553.23, 176.49]]}]\n```" - } - ] - }, - { - "id": 378, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[280.83, 233.56], [298.94, 229.37], [438.9, 228.4], [431.31, 232.44], [280.83, 181.97], [298.94, 181.34], [438.9, 181.19], [431.31, 181.8]]}, {\"category\": \"car\", \"corners_3d\": [[647.36, 228.85], [703.07, 228.55], [739.16, 242.55], [669.59, 243.02], [647.36, 171.8], [703.07, 171.8], [739.16, 171.54], [669.59, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[488.44, 262.85], [407.52, 262.96], [447.75, 244.3], [511.89, 244.23], [488.44, 188.75], [407.52, 188.77], [447.75, 185.47], [511.89, 185.46]]}, {\"category\": \"car\", \"corners_3d\": [[-684.11, 425.73], [-385.88, 363.55], [117.12, 350.65], [-5.31, 403.53], [-684.11, 208.09], [-385.88, 199.43], [117.12, 197.63], [-5.31, 205.0]]}]\n```" - } - ] - }, - { - "id": 379, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[597.14, 191.09], [622.14, 191.06], [626.93, 192.87], [599.44, 192.91], [597.14, 167.04], [622.14, 167.05], [626.93, 166.47], [599.44, 166.46]]}]\n```" - } - ] - }, - { - "id": 380, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[532.74, 195.17], [532.03, 196.01], [456.42, 196.2], [459.88, 195.35], [532.74, 166.33], [532.03, 166.08], [456.42, 166.03], [459.88, 166.27]]}]\n```" - } - ] - }, - { - "id": 381, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[280.83, 233.56], [298.94, 229.37], [438.9, 228.4], [431.31, 232.44], [280.83, 181.97], [298.94, 181.34], [438.9, 181.19], [431.31, 181.8]]}, {\"category\": \"car\", \"corners_3d\": [[647.36, 228.85], [703.07, 228.55], [739.16, 242.55], [669.59, 243.02], [647.36, 171.8], [703.07, 171.8], [739.16, 171.54], [669.59, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[488.44, 262.85], [407.52, 262.96], [447.75, 244.3], [511.89, 244.23], [488.44, 188.75], [407.52, 188.77], [447.75, 185.47], [511.89, 185.46]]}, {\"category\": \"car\", \"corners_3d\": [[-684.11, 425.73], [-385.88, 363.55], [117.12, 350.65], [-5.31, 403.53], [-684.11, 208.09], [-385.88, 199.43], [117.12, 197.63], [-5.31, 205.0]]}, {\"category\": \"truck\", \"corners_3d\": [[597.14, 191.09], [622.14, 191.06], [626.93, 192.87], [599.44, 192.91], [597.14, 167.04], [622.14, 167.05], [626.93, 166.47], [599.44, 166.46]]}]\n```" - } - ] - }, - { - "id": 382, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[532.74, 195.17], [532.03, 196.01], [456.42, 196.2], [459.88, 195.35], [532.74, 166.33], [532.03, 166.08], [456.42, 166.03], [459.88, 166.27]]}, {\"category\": \"car\", \"corners_3d\": [[280.83, 233.56], [298.94, 229.37], [438.9, 228.4], [431.31, 232.44], [280.83, 181.97], [298.94, 181.34], [438.9, 181.19], [431.31, 181.8]]}, {\"category\": \"car\", \"corners_3d\": [[647.36, 228.85], [703.07, 228.55], [739.16, 242.55], [669.59, 243.02], [647.36, 171.8], [703.07, 171.8], [739.16, 171.54], [669.59, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[488.44, 262.85], [407.52, 262.96], [447.75, 244.3], [511.89, 244.23], [488.44, 188.75], [407.52, 188.77], [447.75, 185.47], [511.89, 185.46]]}, {\"category\": \"car\", \"corners_3d\": [[-684.11, 425.73], [-385.88, 363.55], [117.12, 350.65], [-5.31, 403.53], [-684.11, 208.09], [-385.88, 199.43], [117.12, 197.63], [-5.31, 205.0]]}]\n```" - } - ] - }, - { - "id": 383, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[532.74, 195.17], [532.03, 196.01], [456.42, 196.2], [459.88, 195.35], [532.74, 166.33], [532.03, 166.08], [456.42, 166.03], [459.88, 166.27]]}, {\"category\": \"truck\", \"corners_3d\": [[597.14, 191.09], [622.14, 191.06], [626.93, 192.87], [599.44, 192.91], [597.14, 167.04], [622.14, 167.05], [626.93, 166.47], [599.44, 166.46]]}]\n```" - } - ] - }, - { - "id": 384, - "data_source": "Mono3DRefer", - "scene_name": "002845", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[532.74, 195.17], [532.03, 196.01], [456.42, 196.2], [459.88, 195.35], [532.74, 166.33], [532.03, 166.08], [456.42, 166.03], [459.88, 166.27]]}, {\"category\": \"car\", \"corners_3d\": [[280.83, 233.56], [298.94, 229.37], [438.9, 228.4], [431.31, 232.44], [280.83, 181.97], [298.94, 181.34], [438.9, 181.19], [431.31, 181.8]]}, {\"category\": \"car\", \"corners_3d\": [[647.36, 228.85], [703.07, 228.55], [739.16, 242.55], [669.59, 243.02], [647.36, 171.8], [703.07, 171.8], [739.16, 171.54], [669.59, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[488.44, 262.85], [407.52, 262.96], [447.75, 244.3], [511.89, 244.23], [488.44, 188.75], [407.52, 188.77], [447.75, 185.47], [511.89, 185.46]]}, {\"category\": \"car\", \"corners_3d\": [[-684.11, 425.73], [-385.88, 363.55], [117.12, 350.65], [-5.31, 403.53], [-684.11, 208.09], [-385.88, 199.43], [117.12, 197.63], [-5.31, 205.0]]}, {\"category\": \"truck\", \"corners_3d\": [[597.14, 191.09], [622.14, 191.06], [626.93, 192.87], [599.44, 192.91], [597.14, 167.04], [622.14, 167.05], [626.93, 166.47], [599.44, 166.46]]}]\n```" - } - ] - }, - { - "id": 385, - "data_source": "Mono3DRefer", - "scene_name": "002880", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[771.67, 239.09], [697.06, 240.01], [658.34, 225.76], [717.71, 225.18], [771.67, 159.5], [697.06, 159.31], [658.34, 162.19], [717.71, 162.3]]}, {\"category\": \"car\", \"corners_3d\": [[428.5, 248.35], [361.88, 249.24], [391.86, 234.25], [445.25, 233.67], [428.5, 185.78], [361.88, 185.94], [391.86, 183.37], [445.25, 183.27]]}, {\"category\": \"car\", \"corners_3d\": [[468.35, 225.14], [423.17, 225.47], [437.34, 218.9], [476.83, 218.65], [468.35, 188.65], [423.17, 188.75], [437.34, 186.76], [476.83, 186.69]]}]\n```" - } - ] - }, - { - "id": 386, - "data_source": "Mono3DRefer", - "scene_name": "002934", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[562.32, 344.84], [688.53, 329.36], [972.12, 425.2], [803.62, 468.03], [562.32, 203.57], [688.53, 200.81], [972.12, 217.92], [803.62, 225.57]]}, {\"category\": \"car\", \"corners_3d\": [[442.34, 289.12], [536.89, 282.21], [663.79, 329.55], [536.31, 344.14], [442.34, 207.84], [536.89, 205.76], [663.79, 220.0], [536.31, 224.39]]}, {\"category\": \"car\", \"corners_3d\": [[380.86, 272.73], [451.13, 268.21], [512.91, 293.28], [426.78, 300.58], [380.86, 208.58], [451.13, 206.96], [512.91, 215.93], [426.78, 218.54]]}, {\"category\": \"car\", \"corners_3d\": [[328.29, 259.58], [387.92, 256.36], [419.61, 271.76], [349.93, 276.32], [328.29, 203.17], [387.92, 202.04], [419.61, 207.42], [349.93, 209.02]]}, {\"category\": \"car\", \"corners_3d\": [[140.6, 272.76], [74.62, 276.25], [99.85, 260.7], [155.46, 258.16], [140.6, 211.42], [74.62, 212.77], [99.85, 206.77], [155.46, 205.79]]}, {\"category\": \"car\", \"corners_3d\": [[82.26, 319.76], [-27.45, 328.94], [15.04, 297.87], [101.93, 291.91], [82.26, 236.54], [-27.45, 240.52], [15.04, 227.05], [101.93, 224.47]]}, {\"category\": \"car\", \"corners_3d\": [[-33.49, 418.83], [-291.54, 452.22], [-100.9, 342.91], [49.33, 329.93], [-33.49, 255.56], [-291.54, 266.79], [-100.9, 230.04], [49.33, 225.67]]}, {\"category\": \"car\", \"corners_3d\": [[279.72, 244.16], [319.03, 242.33], [337.06, 251.2], [293.12, 253.53], [279.72, 212.74], [319.03, 211.72], [337.06, 216.68], [293.12, 217.99]]}, {\"category\": \"car\", \"corners_3d\": [[179.25, 244.97], [146.82, 246.35], [150.64, 240.62], [180.51, 239.44], [179.25, 218.34], [146.82, 219.22], [150.64, 215.6], [180.51, 214.85]]}]\n```" - } - ] - }, - { - "id": 387, - "data_source": "Mono3DRefer", - "scene_name": "002995", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[658.17, 210.18], [685.11, 210.18], [693.72, 214.47], [663.69, 214.47], [658.17, 186.15], [685.11, 186.15], [693.72, 187.67], [663.69, 187.67]]}]\n```" - } - ] - }, - { - "id": 388, - "data_source": "Mono3DRefer", - "scene_name": "002997", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[483.9, 237.18], [416.73, 237.3], [451.14, 224.83], [505.28, 224.75], [483.9, 175.31], [416.73, 175.32], [451.14, 174.84], [505.28, 174.84]]}]\n```" - } - ] - }, - { - "id": 389, - "data_source": "Mono3DRefer", - "scene_name": "003038", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[646.1, 203.13], [617.84, 202.89], [628.72, 200.51], [654.8, 200.71], [646.1, 177.75], [617.84, 177.71], [628.72, 177.33], [654.8, 177.36]]}]\n```" - } - ] - }, - { - "id": 390, - "data_source": "Mono3DRefer", - "scene_name": "003082", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[479.21, 206.54], [464.37, 206.52], [466.56, 206.06], [481.2, 206.07], [479.21, 177.51], [464.37, 177.5], [466.56, 177.44], [481.2, 177.44]]}]\n```" - } - ] - }, - { - "id": 391, - "data_source": "Mono3DRefer", - "scene_name": "003162", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[785.39, 301.76], [797.75, 310.02], [701.37, 310.02], [695.17, 301.76], [785.39, 160.68], [797.75, 159.32], [701.37, 159.32], [695.17, 160.68]]}]\n```" - } - ] - }, - { - "id": 392, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[412.36, 283.75], [272.03, 283.77], [403.22, 240.58], [488.91, 240.58], [412.36, 71.7], [272.03, 71.68], [403.22, 111.07], [488.91, 111.08]]}]\n```" - } - ] - }, - { - "id": 393, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[597.72, 238.01], [614.7, 236.86], [625.71, 238.13], [608.58, 239.33], [597.72, 165.98], [614.7, 166.1], [625.71, 165.97], [608.58, 165.84]]}]\n```" - } - ] - }, - { - "id": 394, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[617.63, 233.22], [677.89, 233.13], [695.1, 245.74], [622.23, 245.88], [617.63, 175.6], [677.89, 175.6], [695.1, 176.17], [622.23, 176.18]]}]\n```" - } - ] - }, - { - "id": 395, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[412.36, 283.75], [272.03, 283.77], [403.22, 240.58], [488.91, 240.58], [412.36, 71.7], [272.03, 71.68], [403.22, 111.07], [488.91, 111.08]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[597.72, 238.01], [614.7, 236.86], [625.71, 238.13], [608.58, 239.33], [597.72, 165.98], [614.7, 166.1], [625.71, 165.97], [608.58, 165.84]]}]\n```" - } - ] - }, - { - "id": 396, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[412.36, 283.75], [272.03, 283.77], [403.22, 240.58], [488.91, 240.58], [412.36, 71.7], [272.03, 71.68], [403.22, 111.07], [488.91, 111.08]]}, {\"category\": \"car\", \"corners_3d\": [[617.63, 233.22], [677.89, 233.13], [695.1, 245.74], [622.23, 245.88], [617.63, 175.6], [677.89, 175.6], [695.1, 176.17], [622.23, 176.18]]}]\n```" - } - ] - }, - { - "id": 397, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[617.63, 233.22], [677.89, 233.13], [695.1, 245.74], [622.23, 245.88], [617.63, 175.6], [677.89, 175.6], [695.1, 176.17], [622.23, 176.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[597.72, 238.01], [614.7, 236.86], [625.71, 238.13], [608.58, 239.33], [597.72, 165.98], [614.7, 166.1], [625.71, 165.97], [608.58, 165.84]]}]\n```" - } - ] - }, - { - "id": 398, - "data_source": "Mono3DRefer", - "scene_name": "003182", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[412.36, 283.75], [272.03, 283.77], [403.22, 240.58], [488.91, 240.58], [412.36, 71.7], [272.03, 71.68], [403.22, 111.07], [488.91, 111.08]]}, {\"category\": \"car\", \"corners_3d\": [[617.63, 233.22], [677.89, 233.13], [695.1, 245.74], [622.23, 245.88], [617.63, 175.6], [677.89, 175.6], [695.1, 176.17], [622.23, 176.18]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[597.72, 238.01], [614.7, 236.86], [625.71, 238.13], [608.58, 239.33], [597.72, 165.98], [614.7, 166.1], [625.71, 165.97], [608.58, 165.84]]}]\n```" - } - ] - }, - { - "id": 399, - "data_source": "Mono3DRefer", - "scene_name": "003292", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[599.99, 244.07], [567.46, 244.13], [570.29, 236.69], [599.43, 236.64], [599.99, 213.97], [567.46, 214.01], [570.29, 209.72], [599.43, 209.68]]}]\n```" - } - ] - }, - { - "id": 400, - "data_source": "Mono3DRefer", - "scene_name": "003312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[782.69, 310.04], [920.76, 308.74], [1277.61, 448.59], [1001.68, 454.03], [782.69, 149.55], [920.76, 149.78], [1277.61, 126.03], [1001.68, 125.1]]}]\n```" - } - ] - }, - { - "id": 401, - "data_source": "Mono3DRefer", - "scene_name": "003312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[329.87, 439.48], [66.9, 441.49], [324.13, 310.53], [458.41, 310.0], [329.87, 200.55], [66.9, 200.75], [324.13, 187.15], [458.41, 187.1]]}, {\"category\": \"car\", \"corners_3d\": [[693.15, 247.93], [759.74, 247.66], [820.27, 273.03], [731.29, 273.52], [693.15, 177.67], [759.74, 177.65], [820.27, 179.28], [731.29, 179.31]]}, {\"category\": \"car\", \"corners_3d\": [[664.2, 228.84], [715.08, 228.61], [747.53, 240.81], [685.63, 241.16], [664.2, 177.75], [715.08, 177.73], [747.53, 178.8], [685.63, 178.83]]}, {\"category\": \"car\", \"corners_3d\": [[693.05, 193.35], [670.16, 193.39], [662.35, 191.86], [683.54, 191.82], [693.05, 169.37], [670.16, 169.36], [662.35, 169.62], [683.54, 169.63]]}]\n```" - } - ] - }, - { - "id": 402, - "data_source": "Mono3DRefer", - "scene_name": "003312", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[329.87, 439.48], [66.9, 441.49], [324.13, 310.53], [458.41, 310.0], [329.87, 200.55], [66.9, 200.75], [324.13, 187.15], [458.41, 187.1]]}, {\"category\": \"van\", \"corners_3d\": [[782.69, 310.04], [920.76, 308.74], [1277.61, 448.59], [1001.68, 454.03], [782.69, 149.55], [920.76, 149.78], [1277.61, 126.03], [1001.68, 125.1]]}, {\"category\": \"car\", \"corners_3d\": [[693.15, 247.93], [759.74, 247.66], [820.27, 273.03], [731.29, 273.52], [693.15, 177.67], [759.74, 177.65], [820.27, 179.28], [731.29, 179.31]]}, {\"category\": \"car\", \"corners_3d\": [[664.2, 228.84], [715.08, 228.61], [747.53, 240.81], [685.63, 241.16], [664.2, 177.75], [715.08, 177.73], [747.53, 178.8], [685.63, 178.83]]}, {\"category\": \"car\", \"corners_3d\": [[693.05, 193.35], [670.16, 193.39], [662.35, 191.86], [683.54, 191.82], [693.05, 169.37], [670.16, 169.36], [662.35, 169.62], [683.54, 169.63]]}]\n```" - } - ] - }, - { - "id": 403, - "data_source": "Mono3DRefer", - "scene_name": "003448", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[650.43, 219.9], [697.83, 219.78], [723.7, 230.2], [665.84, 230.38], [650.43, 177.37], [697.83, 177.36], [723.7, 178.36], [665.84, 178.37]]}]\n```" - } - ] - }, - { - "id": 404, - "data_source": "Mono3DRefer", - "scene_name": "003467", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[646.98, 224.55], [694.93, 224.42], [717.12, 234.48], [659.85, 234.68], [646.98, 168.8], [694.93, 168.82], [717.12, 168.03], [659.85, 168.01]]}]\n```" - } - ] - }, - { - "id": 405, - "data_source": "Mono3DRefer", - "scene_name": "003467", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[769.23, 300.96], [877.46, 299.81], [1125.68, 401.25], [933.64, 405.01], [769.23, 191.54], [877.46, 191.38], [1125.68, 206.18], [933.64, 206.72]]}]\n```" - } - ] - }, - { - "id": 406, - "data_source": "Mono3DRefer", - "scene_name": "003467", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[769.23, 300.96], [877.46, 299.81], [1125.68, 401.25], [933.64, 405.01], [769.23, 191.54], [877.46, 191.38], [1125.68, 206.18], [933.64, 206.72]]}, {\"category\": \"van\", \"corners_3d\": [[646.98, 224.55], [694.93, 224.42], [717.12, 234.48], [659.85, 234.68], [646.98, 168.8], [694.93, 168.82], [717.12, 168.03], [659.85, 168.01]]}]\n```" - } - ] - }, - { - "id": 407, - "data_source": "Mono3DRefer", - "scene_name": "003497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[321.4, 288.56], [450.44, 288.17], [343.3, 373.21], [117.9, 374.41], [321.4, 188.61], [450.44, 188.56], [343.3, 200.14], [117.9, 200.3]]}, {\"category\": \"car\", \"corners_3d\": [[652.25, 212.98], [699.33, 212.85], [727.22, 221.75], [669.73, 221.94], [652.25, 167.67], [699.33, 167.69], [727.22, 166.54], [669.73, 166.52]]}, {\"category\": \"car\", \"corners_3d\": [[452.4, 229.85], [512.35, 229.71], [490.59, 245.96], [413.38, 246.19], [452.4, 174.4], [512.35, 174.4], [490.59, 174.84], [413.38, 174.85]]}, {\"category\": \"car\", \"corners_3d\": [[719.57, 270.57], [822.24, 270.16], [947.92, 322.48], [790.51, 323.44], [719.57, 166.97], [822.24, 166.99], [947.92, 163.84], [790.51, 163.78]]}, {\"category\": \"car\", \"corners_3d\": [[629.4, 191.17], [658.93, 191.23], [658.06, 193.0], [625.72, 192.92], [629.4, 166.38], [658.93, 166.36], [658.06, 165.74], [625.72, 165.76]]}]\n```" - } - ] - }, - { - "id": 408, - "data_source": "Mono3DRefer", - "scene_name": "003497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[686.48, 175.44], [704.53, 175.47], [685.96, 175.65], [666.93, 175.61], [686.48, 150.62], [704.53, 150.35], [685.96, 148.81], [666.93, 149.12]]}]\n```" - } - ] - }, - { - "id": 409, - "data_source": "Mono3DRefer", - "scene_name": "003497", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[321.4, 288.56], [450.44, 288.17], [343.3, 373.21], [117.9, 374.41], [321.4, 188.61], [450.44, 188.56], [343.3, 200.14], [117.9, 200.3]]}, {\"category\": \"car\", \"corners_3d\": [[652.25, 212.98], [699.33, 212.85], [727.22, 221.75], [669.73, 221.94], [652.25, 167.67], [699.33, 167.69], [727.22, 166.54], [669.73, 166.52]]}, {\"category\": \"car\", \"corners_3d\": [[452.4, 229.85], [512.35, 229.71], [490.59, 245.96], [413.38, 246.19], [452.4, 174.4], [512.35, 174.4], [490.59, 174.84], [413.38, 174.85]]}, {\"category\": \"car\", \"corners_3d\": [[719.57, 270.57], [822.24, 270.16], [947.92, 322.48], [790.51, 323.44], [719.57, 166.97], [822.24, 166.99], [947.92, 163.84], [790.51, 163.78]]}, {\"category\": \"car\", \"corners_3d\": [[629.4, 191.17], [658.93, 191.23], [658.06, 193.0], [625.72, 192.92], [629.4, 166.38], [658.93, 166.36], [658.06, 165.74], [625.72, 165.76]]}, {\"category\": \"van\", \"corners_3d\": [[686.48, 175.44], [704.53, 175.47], [685.96, 175.65], [666.93, 175.61], [686.48, 150.62], [704.53, 150.35], [685.96, 148.81], [666.93, 149.12]]}]\n```" - } - ] - }, - { - "id": 410, - "data_source": "Mono3DRefer", - "scene_name": "003515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[975.56, 345.62], [813.01, 346.44], [731.09, 281.34], [833.1, 281.02], [975.56, 199.56], [813.01, 199.69], [731.09, 189.62], [833.1, 189.58]]}, {\"category\": \"car\", \"corners_3d\": [[388.63, 329.19], [240.56, 329.85], [350.65, 281.03], [452.48, 280.71], [388.63, 202.31], [240.56, 202.43], [350.65, 193.23], [452.48, 193.18]]}, {\"category\": \"car\", \"corners_3d\": [[685.64, 244.55], [752.55, 244.48], [793.54, 264.11], [708.32, 264.21], [685.64, 185.99], [752.55, 185.98], [793.54, 189.58], [708.32, 189.6]]}, {\"category\": \"car\", \"corners_3d\": [[743.43, 239.77], [679.75, 239.83], [666.28, 228.26], [718.97, 228.22], [743.43, 183.2], [679.75, 183.21], [666.28, 181.42], [718.97, 181.41]]}, {\"category\": \"car\", \"corners_3d\": [[729.74, 210.76], [720.23, 208.9], [805.83, 208.45], [819.59, 210.27], [729.74, 175.12], [720.23, 175.0], [805.83, 174.98], [819.59, 175.09]]}]\n```" - } - ] - }, - { - "id": 411, - "data_source": "Mono3DRefer", - "scene_name": "003547", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[481.81, 252.18], [400.9, 252.63], [439.07, 234.93], [501.93, 234.65], [481.81, 181.6], [400.9, 181.65], [439.07, 179.7], [501.93, 179.67]]}, {\"category\": \"car\", \"corners_3d\": [[704.89, 221.07], [660.11, 221.1], [652.37, 214.7], [691.21, 214.67], [704.89, 175.11], [660.11, 175.11], [652.37, 174.81], [691.21, 174.81]]}, {\"category\": \"car\", \"corners_3d\": [[547.22, 210.68], [509.47, 210.58], [526.06, 205.96], [559.22, 206.04], [547.22, 177.63], [509.47, 177.62], [526.06, 177.04], [559.22, 177.05]]}, {\"category\": \"car\", \"corners_3d\": [[637.64, 208.91], [669.24, 208.8], [681.42, 212.79], [646.34, 212.93], [637.64, 179.46], [669.24, 179.44], [681.42, 180.17], [646.34, 180.19]]}]\n```" - } - ] - }, - { - "id": 412, - "data_source": "Mono3DRefer", - "scene_name": "003557", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[439.07, 296.67], [324.77, 296.1], [404.69, 263.89], [489.3, 264.21], [439.07, 194.6], [324.77, 194.5], [404.69, 188.85], [489.3, 188.9]]}, {\"category\": \"car\", \"corners_3d\": [[523.34, 206.68], [548.51, 206.62], [546.46, 209.33], [519.27, 209.4], [523.34, 181.27], [548.51, 181.25], [546.46, 181.93], [519.27, 181.94]]}]\n```" - } - ] - }, - { - "id": 413, - "data_source": "Mono3DRefer", - "scene_name": "003559", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[733.33, 259.98], [803.75, 259.73], [890.23, 294.63], [791.76, 295.12], [733.33, 189.9], [803.75, 189.85], [890.23, 196.68], [791.76, 196.77]]}]\n```" - } - ] - }, - { - "id": 414, - "data_source": "Mono3DRefer", - "scene_name": "003559", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[664.87, 218.63], [705.0, 218.61], [721.13, 225.84], [674.67, 225.87], [664.87, 172.17], [705.0, 172.17], [721.13, 172.07], [674.67, 172.06]]}]\n```" - } - ] - }, - { - "id": 415, - "data_source": "Mono3DRefer", - "scene_name": "003559", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[733.33, 259.98], [803.75, 259.73], [890.23, 294.63], [791.76, 295.12], [733.33, 189.9], [803.75, 189.85], [890.23, 196.68], [791.76, 196.77]]}, {\"category\": \"van\", \"corners_3d\": [[664.87, 218.63], [705.0, 218.61], [721.13, 225.84], [674.67, 225.87], [664.87, 172.17], [705.0, 172.17], [721.13, 172.07], [674.67, 172.06]]}]\n```" - } - ] - }, - { - "id": 416, - "data_source": "Mono3DRefer", - "scene_name": "003580", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[127.86, 256.5], [126.56, 257.74], [49.72, 261.36], [52.25, 260.01], [127.86, 192.47], [126.56, 192.76], [49.72, 193.6], [52.25, 193.29]]}]\n```" - } - ] - }, - { - "id": 417, - "data_source": "Mono3DRefer", - "scene_name": "003580", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[597.89, 204.66], [625.85, 204.68], [625.97, 207.62], [595.43, 207.59], [597.89, 177.03], [625.85, 177.04], [625.97, 177.42], [595.43, 177.42]]}]\n```" - } - ] - }, - { - "id": 418, - "data_source": "Mono3DRefer", - "scene_name": "003580", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[597.89, 204.66], [625.85, 204.68], [625.97, 207.62], [595.43, 207.59], [597.89, 177.03], [625.85, 177.04], [625.97, 177.42], [595.43, 177.42]]}, {\"category\": \"cyclist\", \"corners_3d\": [[127.86, 256.5], [126.56, 257.74], [49.72, 261.36], [52.25, 260.01], [127.86, 192.47], [126.56, 192.76], [49.72, 193.6], [52.25, 193.29]]}]\n```" - } - ] - }, - { - "id": 419, - "data_source": "Mono3DRefer", - "scene_name": "003676", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[714.79, 264.06], [808.52, 263.95], [910.2, 309.0], [770.21, 309.24], [714.79, 167.34], [808.52, 167.35], [910.2, 164.62], [770.21, 164.61]]}, {\"category\": \"car\", \"corners_3d\": [[-590.9, 416.71], [-294.59, 357.64], [95.77, 360.93], [-79.82, 422.48], [-590.9, 194.71], [-294.59, 189.42], [95.77, 189.71], [-79.82, 195.23]]}, {\"category\": \"car\", \"corners_3d\": [[276.26, 309.96], [202.64, 342.35], [-261.12, 345.56], [-96.4, 312.05], [276.26, 187.35], [202.64, 190.77], [-261.12, 191.11], [-96.4, 187.57]]}, {\"category\": \"car\", \"corners_3d\": [[1.11, 298.62], [93.03, 280.92], [331.37, 283.22], [277.19, 301.75], [1.11, 189.01], [93.03, 186.74], [331.37, 187.03], [277.19, 189.41]]}, {\"category\": \"car\", \"corners_3d\": [[126.09, 270.69], [183.13, 259.5], [362.92, 259.98], [328.87, 271.29], [126.09, 182.26], [183.13, 181.19], [362.92, 181.23], [328.87, 182.32]]}, {\"category\": \"car\", \"corners_3d\": [[448.78, 240.97], [431.45, 247.7], [234.64, 247.29], [269.51, 240.63], [448.78, 163.73], [431.45, 162.83], [234.64, 162.88], [269.51, 163.78]]}, {\"category\": \"car\", \"corners_3d\": [[214.38, 260.33], [254.2, 251.7], [402.85, 251.89], [379.25, 260.56], [214.38, 177.34], [254.2, 176.89], [402.85, 176.9], [379.25, 177.35]]}, {\"category\": \"car\", \"corners_3d\": [[688.18, 239.33], [754.33, 239.03], [794.04, 253.6], [713.46, 254.05], [688.18, 181.3], [754.33, 181.26], [794.04, 183.12], [713.46, 183.17]]}, {\"category\": \"car\", \"corners_3d\": [[289.96, 236.96], [314.95, 231.93], [439.15, 231.91], [424.74, 236.94], [289.96, 165.59], [314.95, 166.16], [439.15, 166.16], [424.74, 165.59]]}, {\"category\": \"car\", \"corners_3d\": [[339.39, 225.6], [358.38, 222.18], [480.41, 222.45], [469.82, 225.91], [339.39, 173.45], [358.38, 173.41], [480.41, 173.41], [469.82, 173.45]]}, {\"category\": \"car\", \"corners_3d\": [[645.73, 206.73], [677.89, 206.68], [687.9, 210.48], [652.15, 210.53], [645.73, 177.31], [677.89, 177.3], [687.9, 177.8], [652.15, 177.81]]}, {\"category\": \"car\", \"corners_3d\": [[670.48, 224.42], [721.45, 224.42], [742.24, 234.05], [681.75, 234.05], [670.48, 177.42], [721.45, 177.42], [742.24, 178.27], [681.75, 178.27]]}]\n```" - } - ] - }, - { - "id": 420, - "data_source": "Mono3DRefer", - "scene_name": "003678", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[425.5, 202.73], [398.73, 202.69], [416.19, 200.51], [441.04, 200.54], [425.5, 180.72], [398.73, 180.71], [416.19, 180.13], [441.04, 180.14]]}]\n```" - } - ] - }, - { - "id": 421, - "data_source": "Mono3DRefer", - "scene_name": "003678", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[793.63, 214.96], [809.42, 215.01], [821.16, 218.04], [804.22, 217.99], [793.63, 164.64], [809.42, 164.63], [821.16, 164.04], [804.22, 164.05]]}]\n```" - } - ] - }, - { - "id": 422, - "data_source": "Mono3DRefer", - "scene_name": "003678", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[793.63, 214.96], [809.42, 215.01], [821.16, 218.04], [804.22, 217.99], [793.63, 164.64], [809.42, 164.63], [821.16, 164.04], [804.22, 164.05]]}, {\"category\": \"car\", \"corners_3d\": [[425.5, 202.73], [398.73, 202.69], [416.19, 200.51], [441.04, 200.54], [425.5, 180.72], [398.73, 180.71], [416.19, 180.13], [441.04, 180.14]]}]\n```" - } - ] - }, - { - "id": 423, - "data_source": "Mono3DRefer", - "scene_name": "003755", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1210.56, 444.2], [954.54, 442.38], [801.45, 317.77], [938.13, 318.3], [1210.56, 184.04], [954.54, 183.96], [801.45, 178.83], [938.13, 178.85]]}, {\"category\": \"car\", \"corners_3d\": [[475.47, 282.47], [366.3, 281.98], [442.3, 250.61], [520.22, 250.85], [475.47, 186.7], [366.3, 186.64], [442.3, 182.68], [520.22, 182.71]]}, {\"category\": \"car\", \"corners_3d\": [[712.0, 257.96], [787.46, 257.69], [849.97, 284.35], [750.97, 284.8], [712.0, 187.66], [787.46, 187.62], [849.97, 192.25], [750.97, 192.33]]}, {\"category\": \"car\", \"corners_3d\": [[514.44, 245.39], [445.28, 245.19], [477.62, 232.57], [534.76, 232.71], [514.44, 180.72], [445.28, 180.69], [477.62, 179.33], [534.76, 179.34]]}, {\"category\": \"car\", \"corners_3d\": [[771.37, 249.82], [699.55, 249.75], [681.7, 233.28], [738.11, 233.33], [771.37, 186.84], [699.55, 186.82], [681.7, 183.83], [738.11, 183.84]]}, {\"category\": \"car\", \"corners_3d\": [[555.07, 219.99], [513.62, 220.02], [523.29, 214.85], [560.19, 214.83], [555.07, 179.97], [513.62, 179.98], [523.29, 179.19], [560.19, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[710.62, 213.6], [673.32, 213.56], [667.35, 208.69], [700.18, 208.73], [710.62, 181.28], [673.32, 181.27], [667.35, 180.26], [700.18, 180.27]]}, {\"category\": \"car\", \"corners_3d\": [[653.4, 196.16], [673.99, 196.18], [677.14, 197.92], [655.01, 197.9], [653.4, 176.27], [673.99, 176.28], [677.14, 176.53], [655.01, 176.53]]}]\n```" - } - ] - }, - { - "id": 424, - "data_source": "Mono3DRefer", - "scene_name": "003755", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[647.67, 192.41], [668.77, 192.43], [671.46, 194.2], [648.45, 194.17], [647.67, 167.11], [668.77, 167.1], [671.46, 166.58], [648.45, 166.59]]}]\n```" - } - ] - }, - { - "id": 425, - "data_source": "Mono3DRefer", - "scene_name": "003755", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1210.56, 444.2], [954.54, 442.38], [801.45, 317.77], [938.13, 318.3], [1210.56, 184.04], [954.54, 183.96], [801.45, 178.83], [938.13, 178.85]]}, {\"category\": \"car\", \"corners_3d\": [[475.47, 282.47], [366.3, 281.98], [442.3, 250.61], [520.22, 250.85], [475.47, 186.7], [366.3, 186.64], [442.3, 182.68], [520.22, 182.71]]}, {\"category\": \"car\", \"corners_3d\": [[712.0, 257.96], [787.46, 257.69], [849.97, 284.35], [750.97, 284.8], [712.0, 187.66], [787.46, 187.62], [849.97, 192.25], [750.97, 192.33]]}, {\"category\": \"car\", \"corners_3d\": [[514.44, 245.39], [445.28, 245.19], [477.62, 232.57], [534.76, 232.71], [514.44, 180.72], [445.28, 180.69], [477.62, 179.33], [534.76, 179.34]]}, {\"category\": \"car\", \"corners_3d\": [[771.37, 249.82], [699.55, 249.75], [681.7, 233.28], [738.11, 233.33], [771.37, 186.84], [699.55, 186.82], [681.7, 183.83], [738.11, 183.84]]}, {\"category\": \"car\", \"corners_3d\": [[555.07, 219.99], [513.62, 220.02], [523.29, 214.85], [560.19, 214.83], [555.07, 179.97], [513.62, 179.98], [523.29, 179.19], [560.19, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[710.62, 213.6], [673.32, 213.56], [667.35, 208.69], [700.18, 208.73], [710.62, 181.28], [673.32, 181.27], [667.35, 180.26], [700.18, 180.27]]}, {\"category\": \"car\", \"corners_3d\": [[653.4, 196.16], [673.99, 196.18], [677.14, 197.92], [655.01, 197.9], [653.4, 176.27], [673.99, 176.28], [677.14, 176.53], [655.01, 176.53]]}, {\"category\": \"van\", \"corners_3d\": [[647.67, 192.41], [668.77, 192.43], [671.46, 194.2], [648.45, 194.17], [647.67, 167.11], [668.77, 167.1], [671.46, 166.58], [648.45, 166.59]]}]\n```" - } - ] - }, - { - "id": 426, - "data_source": "Mono3DRefer", - "scene_name": "003828", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[536.07, 230.62], [480.33, 230.35], [505.62, 221.9], [553.24, 222.09], [536.07, 180.81], [480.33, 180.77], [505.62, 179.61], [553.24, 179.64]]}, {\"category\": \"car\", \"corners_3d\": [[553.55, 191.46], [573.6, 191.45], [572.66, 192.62], [551.34, 192.64], [553.55, 171.17], [573.6, 171.17], [572.66, 171.06], [551.34, 171.06]]}]\n```" - } - ] - }, - { - "id": 427, - "data_source": "Mono3DRefer", - "scene_name": "003938", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[565.92, 221.44], [514.0, 221.13], [537.69, 213.98], [582.02, 214.21], [565.92, 166.13], [514.0, 166.17], [537.69, 167.16], [582.02, 167.13]]}, {\"category\": \"car\", \"corners_3d\": [[703.29, 202.65], [745.88, 202.71], [763.82, 207.4], [714.52, 207.33], [703.29, 162.76], [745.88, 162.74], [763.82, 161.15], [714.52, 161.18]]}, {\"category\": \"car\", \"corners_3d\": [[583.59, 209.07], [544.79, 208.95], [556.09, 205.16], [590.83, 205.25], [583.59, 165.6], [544.79, 165.62], [556.09, 166.38], [590.83, 166.36]]}, {\"category\": \"car\", \"corners_3d\": [[408.97, 295.63], [284.77, 293.69], [397.36, 258.74], [486.52, 259.72], [408.97, 181.54], [284.77, 181.4], [397.36, 178.93], [486.52, 179.0]]}, {\"category\": \"car\", \"corners_3d\": [[498.14, 259.17], [407.4, 258.18], [471.93, 237.54], [541.08, 238.1], [498.14, 162.45], [407.4, 162.57], [471.93, 165.06], [541.08, 164.99]]}, {\"category\": \"car\", \"corners_3d\": [[720.24, 211.63], [768.16, 211.69], [793.94, 218.66], [737.38, 218.58], [720.24, 171.17], [768.16, 171.17], [793.94, 170.86], [737.38, 170.87]]}]\n```" - } - ] - }, - { - "id": 428, - "data_source": "Mono3DRefer", - "scene_name": "003938", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[1082.3, 211.3], [1138.64, 215.75], [917.78, 216.01], [884.15, 211.52], [1082.3, 115.68], [1138.64, 109.07], [917.78, 108.67], [884.15, 115.36]]}]\n```" - } - ] - }, - { - "id": 429, - "data_source": "Mono3DRefer", - "scene_name": "003938", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[565.92, 221.44], [514.0, 221.13], [537.69, 213.98], [582.02, 214.21], [565.92, 166.13], [514.0, 166.17], [537.69, 167.16], [582.02, 167.13]]}, {\"category\": \"car\", \"corners_3d\": [[703.29, 202.65], [745.88, 202.71], [763.82, 207.4], [714.52, 207.33], [703.29, 162.76], [745.88, 162.74], [763.82, 161.15], [714.52, 161.18]]}, {\"category\": \"car\", \"corners_3d\": [[583.59, 209.07], [544.79, 208.95], [556.09, 205.16], [590.83, 205.25], [583.59, 165.6], [544.79, 165.62], [556.09, 166.38], [590.83, 166.36]]}, {\"category\": \"car\", \"corners_3d\": [[408.97, 295.63], [284.77, 293.69], [397.36, 258.74], [486.52, 259.72], [408.97, 181.54], [284.77, 181.4], [397.36, 178.93], [486.52, 179.0]]}, {\"category\": \"car\", \"corners_3d\": [[498.14, 259.17], [407.4, 258.18], [471.93, 237.54], [541.08, 238.1], [498.14, 162.45], [407.4, 162.57], [471.93, 165.06], [541.08, 164.99]]}, {\"category\": \"van\", \"corners_3d\": [[1082.3, 211.3], [1138.64, 215.75], [917.78, 216.01], [884.15, 211.52], [1082.3, 115.68], [1138.64, 109.07], [917.78, 108.67], [884.15, 115.36]]}, {\"category\": \"car\", \"corners_3d\": [[720.24, 211.63], [768.16, 211.69], [793.94, 218.66], [737.38, 218.58], [720.24, 171.17], [768.16, 171.17], [793.94, 170.86], [737.38, 170.87]]}]\n```" - } - ] - }, - { - "id": 430, - "data_source": "Mono3DRefer", - "scene_name": "004026", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[648.21, 193.6], [648.17, 192.94], [701.53, 193.01], [703.32, 193.68], [648.21, 173.26], [648.17, 173.25], [701.53, 173.25], [703.32, 173.26]]}, {\"category\": \"car\", \"corners_3d\": [[766.29, 205.28], [775.28, 206.98], [675.2, 207.08], [671.19, 205.37], [766.29, 172.62], [775.28, 172.61], [675.2, 172.61], [671.19, 172.62]]}]\n```" - } - ] - }, - { - "id": 431, - "data_source": "Mono3DRefer", - "scene_name": "004110", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[18.6, 242.9], [74.58, 239.7], [100.21, 245.71], [40.28, 249.52], [18.6, 196.6], [74.58, 195.52], [100.21, 197.55], [40.28, 198.85]]}, {\"category\": \"car\", \"corners_3d\": [[61.1, 254.62], [134.65, 249.65], [174.9, 259.53], [94.15, 265.91], [61.1, 196.78], [134.65, 195.33], [174.9, 198.22], [94.15, 200.09]]}]\n```" - } - ] - }, - { - "id": 432, - "data_source": "Mono3DRefer", - "scene_name": "004130", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[189.86, 434.31], [-58.38, 434.39], [246.09, 315.07], [381.05, 315.05], [189.86, 163.83], [-58.38, 163.83], [246.09, 167.95], [381.05, 167.95]]}]\n```" - } - ] - }, - { - "id": 433, - "data_source": "Mono3DRefer", - "scene_name": "004130", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[531.18, 233.82], [471.89, 233.83], [494.61, 223.72], [544.08, 223.72], [531.18, 184.36], [471.89, 184.36], [494.61, 182.45], [544.08, 182.45]]}]\n```" - } - ] - }, - { - "id": 434, - "data_source": "Mono3DRefer", - "scene_name": "004130", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[189.86, 434.31], [-58.38, 434.39], [246.09, 315.07], [381.05, 315.05], [189.86, 163.83], [-58.38, 163.83], [246.09, 167.95], [381.05, 167.95]]}, {\"category\": \"car\", \"corners_3d\": [[531.18, 233.82], [471.89, 233.83], [494.61, 223.72], [544.08, 223.72], [531.18, 184.36], [471.89, 184.36], [494.61, 182.45], [544.08, 182.45]]}]\n```" - } - ] - }, - { - "id": 435, - "data_source": "Mono3DRefer", - "scene_name": "004162", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[412.23, 307.08], [284.36, 307.57], [369.47, 270.61], [462.12, 270.35], [412.23, 184.57], [284.36, 184.61], [369.47, 181.39], [462.12, 181.37]]}, {\"category\": \"car\", \"corners_3d\": [[828.65, 269.75], [738.39, 269.64], [708.16, 245.7], [776.04, 245.76], [828.65, 191.54], [738.39, 191.52], [708.16, 186.9], [776.04, 186.91]]}, {\"category\": \"car\", \"corners_3d\": [[508.25, 247.01], [428.34, 247.26], [459.53, 232.67], [523.72, 232.5], [508.25, 182.23], [428.34, 182.26], [459.53, 180.41], [523.72, 180.39]]}, {\"category\": \"car\", \"corners_3d\": [[658.6, 208.46], [689.02, 208.49], [696.41, 212.58], [662.49, 212.54], [658.6, 179.07], [689.02, 179.08], [696.41, 179.79], [662.49, 179.79]]}]\n```" - } - ] - }, - { - "id": 436, - "data_source": "Mono3DRefer", - "scene_name": "004162", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[642.99, 199.75], [671.44, 199.74], [680.09, 203.13], [648.04, 203.14], [642.99, 165.51], [671.44, 165.52], [680.09, 164.59], [648.04, 164.59]]}]\n```" - } - ] - }, - { - "id": 437, - "data_source": "Mono3DRefer", - "scene_name": "004162", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[412.23, 307.08], [284.36, 307.57], [369.47, 270.61], [462.12, 270.35], [412.23, 184.57], [284.36, 184.61], [369.47, 181.39], [462.12, 181.37]]}, {\"category\": \"car\", \"corners_3d\": [[828.65, 269.75], [738.39, 269.64], [708.16, 245.7], [776.04, 245.76], [828.65, 191.54], [738.39, 191.52], [708.16, 186.9], [776.04, 186.91]]}, {\"category\": \"car\", \"corners_3d\": [[508.25, 247.01], [428.34, 247.26], [459.53, 232.67], [523.72, 232.5], [508.25, 182.23], [428.34, 182.26], [459.53, 180.41], [523.72, 180.39]]}, {\"category\": \"car\", \"corners_3d\": [[658.6, 208.46], [689.02, 208.49], [696.41, 212.58], [662.49, 212.54], [658.6, 179.07], [689.02, 179.08], [696.41, 179.79], [662.49, 179.79]]}, {\"category\": \"van\", \"corners_3d\": [[642.99, 199.75], [671.44, 199.74], [680.09, 203.13], [648.04, 203.14], [642.99, 165.51], [671.44, 165.52], [680.09, 164.59], [648.04, 164.59]]}]\n```" - } - ] - }, - { - "id": 438, - "data_source": "Mono3DRefer", - "scene_name": "004168", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[373.47, 215.43], [340.43, 215.93], [346.64, 212.04], [376.67, 211.63], [373.47, 187.62], [340.43, 187.79], [346.64, 186.44], [376.67, 186.3]]}]\n```" - } - ] - }, - { - "id": 439, - "data_source": "Mono3DRefer", - "scene_name": "004224", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (motorcyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"motorcyclist\", \"corners_3d\": [[403.61, 359.66], [319.46, 357.89], [392.8, 318.82], [459.6, 319.92], [403.61, 150.39], [319.46, 150.6], [392.8, 155.3], [459.6, 155.17]]}]\n```" - } - ] - }, - { - "id": 440, - "data_source": "Mono3DRefer", - "scene_name": "004224", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[710.56, 159.67], [683.47, 159.76], [688.38, 160.89], [713.14, 160.82], [710.56, 134.64], [683.47, 134.89], [688.38, 138.18], [713.14, 137.97]]}]\n```" - } - ] - }, - { - "id": 441, - "data_source": "Mono3DRefer", - "scene_name": "004224", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (motorcyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"motorcyclist\", \"corners_3d\": [[403.61, 359.66], [319.46, 357.89], [392.8, 318.82], [459.6, 319.92], [403.61, 150.39], [319.46, 150.6], [392.8, 155.3], [459.6, 155.17]]}, {\"category\": \"car\", \"corners_3d\": [[710.56, 159.67], [683.47, 159.76], [688.38, 160.89], [713.14, 160.82], [710.56, 134.64], [683.47, 134.89], [688.38, 138.18], [713.14, 137.97]]}]\n```" - } - ] - }, - { - "id": 442, - "data_source": "Mono3DRefer", - "scene_name": "004249", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[640.36, 195.96], [666.97, 195.96], [673.61, 198.67], [643.89, 198.66], [640.36, 163.98], [666.97, 163.98], [673.61, 162.94], [643.89, 162.94]]}]\n```" - } - ] - }, - { - "id": 443, - "data_source": "Mono3DRefer", - "scene_name": "004249", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[524.84, 231.28], [459.9, 231.44], [481.0, 221.74], [535.16, 221.63], [524.84, 178.57], [459.9, 178.59], [481.0, 177.64], [535.16, 177.63]]}, {\"category\": \"car\", \"corners_3d\": [[653.61, 203.76], [681.38, 203.79], [687.26, 206.99], [656.62, 206.96], [653.61, 176.93], [681.38, 176.94], [687.26, 177.36], [656.62, 177.36]]}, {\"category\": \"car\", \"corners_3d\": [[61.67, 534.67], [-316.18, 532.91], [184.2, 339.68], [359.94, 340.05], [61.67, 179.6], [-316.18, 179.57], [184.2, 175.96], [359.94, 175.97]]}, {\"category\": \"car\", \"corners_3d\": [[1751.33, 730.61], [1228.88, 730.94], [816.36, 359.55], [991.28, 359.51], [1751.33, 271.56], [1228.88, 271.62], [816.36, 205.89], [991.28, 205.89]]}, {\"category\": \"car\", \"corners_3d\": [[464.21, 268.83], [373.86, 268.96], [422.04, 248.64], [493.26, 248.56], [464.21, 181.78], [373.86, 181.79], [422.04, 179.9], [493.26, 179.89]]}, {\"category\": \"car\", \"corners_3d\": [[779.03, 247.95], [708.69, 247.88], [689.82, 232.57], [745.78, 232.62], [779.03, 186.98], [708.69, 186.96], [689.82, 184.09], [745.78, 184.09]]}]\n```" - } - ] - }, - { - "id": 444, - "data_source": "Mono3DRefer", - "scene_name": "004249", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[524.84, 231.28], [459.9, 231.44], [481.0, 221.74], [535.16, 221.63], [524.84, 178.57], [459.9, 178.59], [481.0, 177.64], [535.16, 177.63]]}, {\"category\": \"car\", \"corners_3d\": [[653.61, 203.76], [681.38, 203.79], [687.26, 206.99], [656.62, 206.96], [653.61, 176.93], [681.38, 176.94], [687.26, 177.36], [656.62, 177.36]]}, {\"category\": \"van\", \"corners_3d\": [[640.36, 195.96], [666.97, 195.96], [673.61, 198.67], [643.89, 198.66], [640.36, 163.98], [666.97, 163.98], [673.61, 162.94], [643.89, 162.94]]}, {\"category\": \"car\", \"corners_3d\": [[61.67, 534.67], [-316.18, 532.91], [184.2, 339.68], [359.94, 340.05], [61.67, 179.6], [-316.18, 179.57], [184.2, 175.96], [359.94, 175.97]]}, {\"category\": \"car\", \"corners_3d\": [[1751.33, 730.61], [1228.88, 730.94], [816.36, 359.55], [991.28, 359.51], [1751.33, 271.56], [1228.88, 271.62], [816.36, 205.89], [991.28, 205.89]]}, {\"category\": \"car\", \"corners_3d\": [[464.21, 268.83], [373.86, 268.96], [422.04, 248.64], [493.26, 248.56], [464.21, 181.78], [373.86, 181.79], [422.04, 179.9], [493.26, 179.89]]}, {\"category\": \"car\", \"corners_3d\": [[779.03, 247.95], [708.69, 247.88], [689.82, 232.57], [745.78, 232.62], [779.03, 186.98], [708.69, 186.96], [689.82, 184.09], [745.78, 184.09]]}]\n```" - } - ] - }, - { - "id": 445, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[590.42, 368.97], [578.02, 343.84], [687.76, 340.32], [716.41, 364.3], [590.42, 151.46], [578.02, 155.49], [687.76, 156.05], [716.41, 152.21]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[335.31, 396.82], [336.5, 370.44], [453.31, 360.78], [467.56, 384.36], [335.31, 178.65], [336.5, 179.0], [453.31, 179.13], [467.56, 178.82]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1035.42, 356.97], [1131.18, 388.3], [1013.29, 393.64], [933.69, 360.79], [1035.42, 180.14], [1131.18, 179.89], [1013.29, 179.85], [933.69, 180.11]]}]\n```" - } - ] - }, - { - "id": 446, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1159.23, 241.0], [1261.16, 249.01], [1058.29, 256.08], [973.31, 246.43], [1159.23, 152.14], [1261.16, 148.19], [1058.29, 144.69], [973.31, 149.46]]}]\n```" - } - ] - }, - { - "id": 447, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[977.39, 207.87], [1046.76, 210.62], [721.2, 217.63], [674.41, 213.49], [977.39, 110.91], [1046.76, 103.54], [721.2, 84.76], [674.41, 95.87]]}]\n```" - } - ] - }, - { - "id": 448, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[590.42, 368.97], [578.02, 343.84], [687.76, 340.32], [716.41, 364.3], [590.42, 151.46], [578.02, 155.49], [687.76, 156.05], [716.41, 152.21]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[335.31, 396.82], [336.5, 370.44], [453.31, 360.78], [467.56, 384.36], [335.31, 178.65], [336.5, 179.0], [453.31, 179.13], [467.56, 178.82]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1035.42, 356.97], [1131.18, 388.3], [1013.29, 393.64], [933.69, 360.79], [1035.42, 180.14], [1131.18, 179.89], [1013.29, 179.85], [933.69, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[1159.23, 241.0], [1261.16, 249.01], [1058.29, 256.08], [973.31, 246.43], [1159.23, 152.14], [1261.16, 148.19], [1058.29, 144.69], [973.31, 149.46]]}]\n```" - } - ] - }, - { - "id": 449, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[977.39, 207.87], [1046.76, 210.62], [721.2, 217.63], [674.41, 213.49], [977.39, 110.91], [1046.76, 103.54], [721.2, 84.76], [674.41, 95.87]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[590.42, 368.97], [578.02, 343.84], [687.76, 340.32], [716.41, 364.3], [590.42, 151.46], [578.02, 155.49], [687.76, 156.05], [716.41, 152.21]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[335.31, 396.82], [336.5, 370.44], [453.31, 360.78], [467.56, 384.36], [335.31, 178.65], [336.5, 179.0], [453.31, 179.13], [467.56, 178.82]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1035.42, 356.97], [1131.18, 388.3], [1013.29, 393.64], [933.69, 360.79], [1035.42, 180.14], [1131.18, 179.89], [1013.29, 179.85], [933.69, 180.11]]}]\n```" - } - ] - }, - { - "id": 450, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[977.39, 207.87], [1046.76, 210.62], [721.2, 217.63], [674.41, 213.49], [977.39, 110.91], [1046.76, 103.54], [721.2, 84.76], [674.41, 95.87]]}, {\"category\": \"car\", \"corners_3d\": [[1159.23, 241.0], [1261.16, 249.01], [1058.29, 256.08], [973.31, 246.43], [1159.23, 152.14], [1261.16, 148.19], [1058.29, 144.69], [973.31, 149.46]]}]\n```" - } - ] - }, - { - "id": 451, - "data_source": "Mono3DRefer", - "scene_name": "004271", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[977.39, 207.87], [1046.76, 210.62], [721.2, 217.63], [674.41, 213.49], [977.39, 110.91], [1046.76, 103.54], [721.2, 84.76], [674.41, 95.87]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[590.42, 368.97], [578.02, 343.84], [687.76, 340.32], [716.41, 364.3], [590.42, 151.46], [578.02, 155.49], [687.76, 156.05], [716.41, 152.21]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[335.31, 396.82], [336.5, 370.44], [453.31, 360.78], [467.56, 384.36], [335.31, 178.65], [336.5, 179.0], [453.31, 179.13], [467.56, 178.82]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1035.42, 356.97], [1131.18, 388.3], [1013.29, 393.64], [933.69, 360.79], [1035.42, 180.14], [1131.18, 179.89], [1013.29, 179.85], [933.69, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[1159.23, 241.0], [1261.16, 249.01], [1058.29, 256.08], [973.31, 246.43], [1159.23, 152.14], [1261.16, 148.19], [1058.29, 144.69], [973.31, 149.46]]}]\n```" - } - ] - }, - { - "id": 452, - "data_source": "Mono3DRefer", - "scene_name": "004318", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[582.4, 194.43], [614.64, 194.38], [624.74, 195.37], [589.06, 195.43], [582.4, 164.88], [614.64, 164.98], [624.74, 162.8], [589.06, 162.67]]}, {\"category\": \"car\", \"corners_3d\": [[512.78, 196.96], [543.94, 196.88], [549.34, 198.09], [514.98, 198.19], [512.78, 169.24], [543.94, 169.35], [549.34, 167.7], [514.98, 167.57]]}, {\"category\": \"car\", \"corners_3d\": [[506.24, 188.31], [530.93, 188.29], [534.15, 188.54], [507.44, 188.56], [506.24, 167.43], [530.93, 167.53], [534.15, 166.07], [507.44, 165.96]]}]\n```" - } - ] - }, - { - "id": 453, - "data_source": "Mono3DRefer", - "scene_name": "004323", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[359.96, 313.8], [334.92, 328.31], [219.72, 328.79], [256.18, 314.19], [359.96, 150.59], [334.92, 147.34], [219.72, 147.23], [256.18, 150.5]]}]\n```" - } - ] - }, - { - "id": 454, - "data_source": "Mono3DRefer", - "scene_name": "004363", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[293.19, 221.52], [243.09, 222.51], [260.52, 215.68], [303.55, 214.95], [293.19, 164.51], [243.09, 164.35], [260.52, 165.52], [303.55, 165.64]]}]\n```" - } - ] - }, - { - "id": 455, - "data_source": "Mono3DRefer", - "scene_name": "004363", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[200.89, 361.83], [-52.37, 385.01], [94.49, 295.39], [238.12, 287.28], [200.89, 151.93], [-52.37, 149.37], [94.49, 159.29], [238.12, 160.18]]}, {\"category\": \"car\", \"corners_3d\": [[240.61, 231.86], [288.33, 230.5], [285.95, 238.67], [231.22, 240.44], [240.61, 188.77], [288.33, 188.41], [285.95, 190.61], [231.22, 191.09]]}, {\"category\": \"car\", \"corners_3d\": [[425.62, 229.51], [467.16, 228.3], [485.56, 234.94], [439.37, 236.47], [425.62, 187.7], [467.16, 187.38], [485.56, 189.12], [439.37, 189.52]]}, {\"category\": \"car\", \"corners_3d\": [[405.17, 222.56], [442.82, 221.59], [458.82, 227.25], [417.06, 228.47], [405.17, 186.51], [442.82, 186.24], [458.82, 187.8], [417.06, 188.14]]}, {\"category\": \"car\", \"corners_3d\": [[388.37, 211.71], [415.98, 211.29], [418.54, 214.68], [388.49, 215.18], [388.37, 187.36], [415.98, 187.2], [418.54, 188.47], [388.49, 188.65]]}, {\"category\": \"car\", \"corners_3d\": [[451.96, 239.49], [511.7, 237.73], [539.52, 249.88], [469.14, 252.36], [451.96, 184.42], [511.7, 184.11], [539.52, 186.22], [469.14, 186.65]]}]\n```" - } - ] - }, - { - "id": 456, - "data_source": "Mono3DRefer", - "scene_name": "004363", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[200.89, 361.83], [-52.37, 385.01], [94.49, 295.39], [238.12, 287.28], [200.89, 151.93], [-52.37, 149.37], [94.49, 159.29], [238.12, 160.18]]}, {\"category\": \"car\", \"corners_3d\": [[240.61, 231.86], [288.33, 230.5], [285.95, 238.67], [231.22, 240.44], [240.61, 188.77], [288.33, 188.41], [285.95, 190.61], [231.22, 191.09]]}, {\"category\": \"car\", \"corners_3d\": [[425.62, 229.51], [467.16, 228.3], [485.56, 234.94], [439.37, 236.47], [425.62, 187.7], [467.16, 187.38], [485.56, 189.12], [439.37, 189.52]]}, {\"category\": \"car\", \"corners_3d\": [[405.17, 222.56], [442.82, 221.59], [458.82, 227.25], [417.06, 228.47], [405.17, 186.51], [442.82, 186.24], [458.82, 187.8], [417.06, 188.14]]}, {\"category\": \"truck\", \"corners_3d\": [[293.19, 221.52], [243.09, 222.51], [260.52, 215.68], [303.55, 214.95], [293.19, 164.51], [243.09, 164.35], [260.52, 165.52], [303.55, 165.64]]}, {\"category\": \"car\", \"corners_3d\": [[388.37, 211.71], [415.98, 211.29], [418.54, 214.68], [388.49, 215.18], [388.37, 187.36], [415.98, 187.2], [418.54, 188.47], [388.49, 188.65]]}, {\"category\": \"car\", \"corners_3d\": [[451.96, 239.49], [511.7, 237.73], [539.52, 249.88], [469.14, 252.36], [451.96, 184.42], [511.7, 184.11], [539.52, 186.22], [469.14, 186.65]]}]\n```" - } - ] - }, - { - "id": 457, - "data_source": "Mono3DRefer", - "scene_name": "004424", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-133.37, 316.08], [-28.67, 303.12], [63.81, 322.64], [-47.67, 340.02], [-133.37, 241.8], [-28.67, 235.57], [63.81, 244.96], [-47.67, 253.33]]}, {\"category\": \"car\", \"corners_3d\": [[1131.18, 224.08], [1219.5, 226.54], [1279.88, 241.62], [1164.7, 237.63], [1131.18, 166.72], [1219.5, 166.43], [1279.88, 164.62], [1164.7, 165.1]]}, {\"category\": \"car\", \"corners_3d\": [[1086.81, 206.42], [1142.41, 207.57], [1155.45, 213.25], [1090.6, 211.7], [1086.81, 159.52], [1142.41, 159.06], [1155.45, 156.8], [1090.6, 157.42]]}, {\"category\": \"car\", \"corners_3d\": [[137.2, 342.27], [257.01, 319.38], [447.77, 350.34], [335.01, 385.05], [137.2, 221.35], [257.01, 214.8], [447.77, 223.66], [335.01, 233.6]]}, {\"category\": \"car\", \"corners_3d\": [[886.26, 245.68], [971.65, 249.43], [967.65, 270.93], [860.0, 264.87], [886.26, 175.89], [971.65, 176.05], [967.65, 176.94], [860.0, 176.69]]}, {\"category\": \"car\", \"corners_3d\": [[1183.62, 249.09], [1300.46, 252.92], [1468.92, 287.18], [1294.15, 279.53], [1183.62, 161.23], [1300.46, 160.65], [1468.92, 155.42], [1294.15, 156.59]]}, {\"category\": \"car\", \"corners_3d\": [[-27.81, 323.13], [92.9, 306.2], [237.34, 334.42], [109.89, 359.95], [-27.81, 236.98], [92.9, 229.75], [237.34, 241.79], [109.89, 252.69]]}]\n```" - } - ] - }, - { - "id": 458, - "data_source": "Mono3DRefer", - "scene_name": "004438", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[533.87, 236.88], [486.13, 236.96], [501.94, 227.61], [542.71, 227.55], [533.87, 192.82], [486.13, 192.84], [501.94, 189.93], [542.71, 189.91]]}]\n```" - } - ] - }, - { - "id": 459, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[417.6, 275.57], [312.19, 276.26], [378.69, 251.22], [456.04, 250.85], [417.6, 178.32], [312.19, 178.3], [378.69, 179.15], [456.04, 179.16]]}]\n```" - } - ] - }, - { - "id": 460, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[695.92, 220.64], [660.29, 220.84], [655.85, 219.53], [690.32, 219.34], [695.92, 158.97], [660.29, 158.86], [655.85, 159.62], [690.32, 159.72]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[530.28, 213.89], [551.76, 213.8], [552.25, 214.81], [530.09, 214.91], [530.28, 169.73], [551.76, 169.76], [552.25, 169.4], [530.09, 169.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[541.49, 213.26], [564.3, 213.11], [566.76, 214.2], [543.18, 214.36], [541.49, 167.23], [564.3, 167.3], [566.76, 166.81], [543.18, 166.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[164.24, 425.98], [276.61, 399.33], [351.55, 429.35], [232.11, 464.44], [164.24, 161.77], [276.61, 163.92], [351.55, 161.5], [232.11, 158.67]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[269.86, 378.68], [347.93, 365.88], [390.77, 389.14], [305.44, 405.52], [269.86, 172.75], [347.93, 173.32], [390.77, 172.28], [305.44, 171.56]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[124.3, 368.1], [212.29, 357.2], [226.92, 379.93], [127.78, 393.95], [124.3, 172.18], [212.29, 172.72], [226.92, 171.58], [127.78, 170.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[99.95, 324.35], [178.46, 319.48], [163.3, 334.9], [75.07, 340.94], [99.95, 177.96], [178.46, 178.08], [163.3, 177.7], [75.07, 177.55]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[338.29, 318.24], [396.73, 307.95], [447.51, 315.99], [389.45, 327.68], [338.29, 162.33], [396.73, 163.77], [447.51, 162.64], [389.45, 161.0]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[151.81, 336.92], [225.31, 327.44], [251.84, 340.07], [173.41, 351.32], [151.81, 181.37], [225.31, 181.38], [251.84, 181.37], [173.41, 181.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[309.47, 313.08], [368.98, 308.03], [385.79, 319.9], [321.19, 325.97], [309.47, 168.8], [368.98, 169.29], [385.79, 168.15], [321.19, 167.56]]}]\n```" - } - ] - }, - { - "id": 461, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[510.91, 233.81], [479.86, 234.1], [482.23, 229.32], [510.45, 229.08], [510.91, 164.0], [479.86, 163.9], [482.23, 165.51], [510.45, 165.59]]}]\n```" - } - ] - }, - { - "id": 462, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[695.92, 220.64], [660.29, 220.84], [655.85, 219.53], [690.32, 219.34], [695.92, 158.97], [660.29, 158.86], [655.85, 159.62], [690.32, 159.72]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[530.28, 213.89], [551.76, 213.8], [552.25, 214.81], [530.09, 214.91], [530.28, 169.73], [551.76, 169.76], [552.25, 169.4], [530.09, 169.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[541.49, 213.26], [564.3, 213.11], [566.76, 214.2], [543.18, 214.36], [541.49, 167.23], [564.3, 167.3], [566.76, 166.81], [543.18, 166.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[164.24, 425.98], [276.61, 399.33], [351.55, 429.35], [232.11, 464.44], [164.24, 161.77], [276.61, 163.92], [351.55, 161.5], [232.11, 158.67]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[269.86, 378.68], [347.93, 365.88], [390.77, 389.14], [305.44, 405.52], [269.86, 172.75], [347.93, 173.32], [390.77, 172.28], [305.44, 171.56]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[124.3, 368.1], [212.29, 357.2], [226.92, 379.93], [127.78, 393.95], [124.3, 172.18], [212.29, 172.72], [226.92, 171.58], [127.78, 170.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[99.95, 324.35], [178.46, 319.48], [163.3, 334.9], [75.07, 340.94], [99.95, 177.96], [178.46, 178.08], [163.3, 177.7], [75.07, 177.55]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[338.29, 318.24], [396.73, 307.95], [447.51, 315.99], [389.45, 327.68], [338.29, 162.33], [396.73, 163.77], [447.51, 162.64], [389.45, 161.0]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[151.81, 336.92], [225.31, 327.44], [251.84, 340.07], [173.41, 351.32], [151.81, 181.37], [225.31, 181.38], [251.84, 181.37], [173.41, 181.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[309.47, 313.08], [368.98, 308.03], [385.79, 319.9], [321.19, 325.97], [309.47, 168.8], [368.98, 169.29], [385.79, 168.15], [321.19, 167.56]]}, {\"category\": \"car\", \"corners_3d\": [[417.6, 275.57], [312.19, 276.26], [378.69, 251.22], [456.04, 250.85], [417.6, 178.32], [312.19, 178.3], [378.69, 179.15], [456.04, 179.16]]}]\n```" - } - ] - }, - { - "id": 463, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[417.6, 275.57], [312.19, 276.26], [378.69, 251.22], [456.04, 250.85], [417.6, 178.32], [312.19, 178.3], [378.69, 179.15], [456.04, 179.16]]}, {\"category\": \"cyclist\", \"corners_3d\": [[510.91, 233.81], [479.86, 234.1], [482.23, 229.32], [510.45, 229.08], [510.91, 164.0], [479.86, 163.9], [482.23, 165.51], [510.45, 165.59]]}]\n```" - } - ] - }, - { - "id": 464, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[695.92, 220.64], [660.29, 220.84], [655.85, 219.53], [690.32, 219.34], [695.92, 158.97], [660.29, 158.86], [655.85, 159.62], [690.32, 159.72]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[530.28, 213.89], [551.76, 213.8], [552.25, 214.81], [530.09, 214.91], [530.28, 169.73], [551.76, 169.76], [552.25, 169.4], [530.09, 169.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[541.49, 213.26], [564.3, 213.11], [566.76, 214.2], [543.18, 214.36], [541.49, 167.23], [564.3, 167.3], [566.76, 166.81], [543.18, 166.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[164.24, 425.98], [276.61, 399.33], [351.55, 429.35], [232.11, 464.44], [164.24, 161.77], [276.61, 163.92], [351.55, 161.5], [232.11, 158.67]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[269.86, 378.68], [347.93, 365.88], [390.77, 389.14], [305.44, 405.52], [269.86, 172.75], [347.93, 173.32], [390.77, 172.28], [305.44, 171.56]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[124.3, 368.1], [212.29, 357.2], [226.92, 379.93], [127.78, 393.95], [124.3, 172.18], [212.29, 172.72], [226.92, 171.58], [127.78, 170.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[99.95, 324.35], [178.46, 319.48], [163.3, 334.9], [75.07, 340.94], [99.95, 177.96], [178.46, 178.08], [163.3, 177.7], [75.07, 177.55]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[338.29, 318.24], [396.73, 307.95], [447.51, 315.99], [389.45, 327.68], [338.29, 162.33], [396.73, 163.77], [447.51, 162.64], [389.45, 161.0]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[151.81, 336.92], [225.31, 327.44], [251.84, 340.07], [173.41, 351.32], [151.81, 181.37], [225.31, 181.38], [251.84, 181.37], [173.41, 181.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[309.47, 313.08], [368.98, 308.03], [385.79, 319.9], [321.19, 325.97], [309.47, 168.8], [368.98, 169.29], [385.79, 168.15], [321.19, 167.56]]}, {\"category\": \"cyclist\", \"corners_3d\": [[510.91, 233.81], [479.86, 234.1], [482.23, 229.32], [510.45, 229.08], [510.91, 164.0], [479.86, 163.9], [482.23, 165.51], [510.45, 165.59]]}]\n```" - } - ] - }, - { - "id": 465, - "data_source": "Mono3DRefer", - "scene_name": "004474", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[695.92, 220.64], [660.29, 220.84], [655.85, 219.53], [690.32, 219.34], [695.92, 158.97], [660.29, 158.86], [655.85, 159.62], [690.32, 159.72]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[530.28, 213.89], [551.76, 213.8], [552.25, 214.81], [530.09, 214.91], [530.28, 169.73], [551.76, 169.76], [552.25, 169.4], [530.09, 169.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[541.49, 213.26], [564.3, 213.11], [566.76, 214.2], [543.18, 214.36], [541.49, 167.23], [564.3, 167.3], [566.76, 166.81], [543.18, 166.74]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[164.24, 425.98], [276.61, 399.33], [351.55, 429.35], [232.11, 464.44], [164.24, 161.77], [276.61, 163.92], [351.55, 161.5], [232.11, 158.67]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[269.86, 378.68], [347.93, 365.88], [390.77, 389.14], [305.44, 405.52], [269.86, 172.75], [347.93, 173.32], [390.77, 172.28], [305.44, 171.56]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[124.3, 368.1], [212.29, 357.2], [226.92, 379.93], [127.78, 393.95], [124.3, 172.18], [212.29, 172.72], [226.92, 171.58], [127.78, 170.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[99.95, 324.35], [178.46, 319.48], [163.3, 334.9], [75.07, 340.94], [99.95, 177.96], [178.46, 178.08], [163.3, 177.7], [75.07, 177.55]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[338.29, 318.24], [396.73, 307.95], [447.51, 315.99], [389.45, 327.68], [338.29, 162.33], [396.73, 163.77], [447.51, 162.64], [389.45, 161.0]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[151.81, 336.92], [225.31, 327.44], [251.84, 340.07], [173.41, 351.32], [151.81, 181.37], [225.31, 181.38], [251.84, 181.37], [173.41, 181.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[309.47, 313.08], [368.98, 308.03], [385.79, 319.9], [321.19, 325.97], [309.47, 168.8], [368.98, 169.29], [385.79, 168.15], [321.19, 167.56]]}, {\"category\": \"car\", \"corners_3d\": [[417.6, 275.57], [312.19, 276.26], [378.69, 251.22], [456.04, 250.85], [417.6, 178.32], [312.19, 178.3], [378.69, 179.15], [456.04, 179.16]]}, {\"category\": \"cyclist\", \"corners_3d\": [[510.91, 233.81], [479.86, 234.1], [482.23, 229.32], [510.45, 229.08], [510.91, 164.0], [479.86, 163.9], [482.23, 165.51], [510.45, 165.59]]}]\n```" - } - ] - }, - { - "id": 466, - "data_source": "Mono3DRefer", - "scene_name": "004482", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[519.06, 301.94], [524.31, 294.54], [605.75, 294.56], [605.78, 301.97], [519.06, 142.95], [524.31, 145.24], [605.75, 145.23], [605.78, 142.94]]}]\n```" - } - ] - }, - { - "id": 467, - "data_source": "Mono3DRefer", - "scene_name": "004491", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[624.79, 206.35], [590.37, 206.24], [598.37, 202.38], [628.82, 202.47], [624.79, 175.39], [590.37, 175.38], [598.37, 175.09], [628.82, 175.1]]}, {\"category\": \"car\", \"corners_3d\": [[390.99, 222.81], [406.06, 219.38], [531.52, 219.4], [525.68, 222.82], [390.99, 176.32], [406.06, 176.08], [531.52, 176.08], [525.68, 176.32]]}, {\"category\": \"car\", \"corners_3d\": [[1007.89, 350.75], [835.41, 349.12], [765.85, 287.05], [876.81, 287.73], [1007.89, 192.94], [835.41, 192.75], [765.85, 185.75], [876.81, 185.82]]}]\n```" - } - ] - }, - { - "id": 468, - "data_source": "Mono3DRefer", - "scene_name": "004491", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[597.48, 219.11], [545.96, 218.78], [564.42, 212.52], [608.98, 212.77], [597.48, 166.6], [545.96, 166.64], [564.42, 167.49], [608.98, 167.46]]}, {\"category\": \"van\", \"corners_3d\": [[702.88, 213.68], [744.25, 213.78], [760.19, 219.97], [712.53, 219.85], [702.88, 169.06], [744.25, 169.06], [760.19, 168.48], [712.53, 168.49]]}]\n```" - } - ] - }, - { - "id": 469, - "data_source": "Mono3DRefer", - "scene_name": "004491", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[624.79, 206.35], [590.37, 206.24], [598.37, 202.38], [628.82, 202.47], [624.79, 175.39], [590.37, 175.38], [598.37, 175.09], [628.82, 175.1]]}, {\"category\": \"car\", \"corners_3d\": [[390.99, 222.81], [406.06, 219.38], [531.52, 219.4], [525.68, 222.82], [390.99, 176.32], [406.06, 176.08], [531.52, 176.08], [525.68, 176.32]]}, {\"category\": \"car\", \"corners_3d\": [[1007.89, 350.75], [835.41, 349.12], [765.85, 287.05], [876.81, 287.73], [1007.89, 192.94], [835.41, 192.75], [765.85, 185.75], [876.81, 185.82]]}, {\"category\": \"van\", \"corners_3d\": [[597.48, 219.11], [545.96, 218.78], [564.42, 212.52], [608.98, 212.77], [597.48, 166.6], [545.96, 166.64], [564.42, 167.49], [608.98, 167.46]]}, {\"category\": \"van\", \"corners_3d\": [[702.88, 213.68], [744.25, 213.78], [760.19, 219.97], [712.53, 219.85], [702.88, 169.06], [744.25, 169.06], [760.19, 168.48], [712.53, 168.49]]}]\n```" - } - ] - }, - { - "id": 470, - "data_source": "Mono3DRefer", - "scene_name": "004608", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[592.0, 243.79], [558.35, 243.86], [562.24, 236.22], [592.27, 236.16], [592.0, 212.67], [558.35, 212.71], [562.24, 208.42], [592.27, 208.39]]}]\n```" - } - ] - }, - { - "id": 471, - "data_source": "Mono3DRefer", - "scene_name": "004638", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[314.11, 198.26], [279.42, 198.31], [308.92, 195.82], [340.17, 195.78], [314.11, 171.86], [279.42, 171.86], [308.92, 171.96], [340.17, 171.96]]}]\n```" - } - ] - }, - { - "id": 472, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[597.0, 192.22], [597.49, 191.52], [673.03, 191.52], [675.38, 192.22], [597.0, 152.14], [597.49, 152.89], [673.03, 152.89], [675.38, 152.13]]}]\n```" - } - ] - }, - { - "id": 473, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[689.01, 204.32], [678.94, 204.7], [667.34, 204.13], [677.37, 203.76], [689.01, 170.22], [678.94, 170.19], [667.34, 170.24], [677.37, 170.27]]}]\n```" - } - ] - }, - { - "id": 474, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[80.31, 365.72], [-99.28, 365.76], [136.02, 301.67], [255.93, 301.65], [80.31, 203.1], [-99.28, 203.1], [136.02, 193.05], [255.93, 193.05]]}, {\"category\": \"car\", \"corners_3d\": [[901.34, 386.49], [1110.08, 381.42], [2274.13, 816.67], [1687.34, 867.54], [901.34, 189.95], [1110.08, 189.55], [2274.13, 224.39], [1687.34, 228.46]]}, {\"category\": \"car\", \"corners_3d\": [[290.42, 287.3], [181.25, 287.49], [287.94, 258.4], [369.32, 258.3], [290.42, 188.14], [181.25, 188.16], [287.94, 184.28], [369.32, 184.26]]}, {\"category\": \"car\", \"corners_3d\": [[767.72, 278.89], [872.83, 278.28], [1012.59, 328.81], [857.86, 330.15], [767.72, 180.12], [872.83, 180.08], [1012.59, 183.54], [857.86, 183.64]]}, {\"category\": \"car\", \"corners_3d\": [[527.69, 203.08], [503.12, 203.84], [471.79, 201.81], [495.51, 201.14], [527.69, 177.71], [503.12, 177.83], [471.79, 177.5], [495.51, 177.4]]}, {\"category\": \"car\", \"corners_3d\": [[674.1, 191.19], [667.33, 190.62], [721.85, 190.33], [730.21, 190.87], [674.1, 172.6], [667.33, 172.61], [721.85, 172.61], [730.21, 172.6]]}]\n```" - } - ] - }, - { - "id": 475, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[689.01, 204.32], [678.94, 204.7], [667.34, 204.13], [677.37, 203.76], [689.01, 170.22], [678.94, 170.19], [667.34, 170.24], [677.37, 170.27]]}, {\"category\": \"truck\", \"corners_3d\": [[597.0, 192.22], [597.49, 191.52], [673.03, 191.52], [675.38, 192.22], [597.0, 152.14], [597.49, 152.89], [673.03, 152.89], [675.38, 152.13]]}]\n```" - } - ] - }, - { - "id": 476, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[80.31, 365.72], [-99.28, 365.76], [136.02, 301.67], [255.93, 301.65], [80.31, 203.1], [-99.28, 203.1], [136.02, 193.05], [255.93, 193.05]]}, {\"category\": \"car\", \"corners_3d\": [[901.34, 386.49], [1110.08, 381.42], [2274.13, 816.67], [1687.34, 867.54], [901.34, 189.95], [1110.08, 189.55], [2274.13, 224.39], [1687.34, 228.46]]}, {\"category\": \"car\", \"corners_3d\": [[290.42, 287.3], [181.25, 287.49], [287.94, 258.4], [369.32, 258.3], [290.42, 188.14], [181.25, 188.16], [287.94, 184.28], [369.32, 184.26]]}, {\"category\": \"car\", \"corners_3d\": [[767.72, 278.89], [872.83, 278.28], [1012.59, 328.81], [857.86, 330.15], [767.72, 180.12], [872.83, 180.08], [1012.59, 183.54], [857.86, 183.64]]}, {\"category\": \"car\", \"corners_3d\": [[527.69, 203.08], [503.12, 203.84], [471.79, 201.81], [495.51, 201.14], [527.69, 177.71], [503.12, 177.83], [471.79, 177.5], [495.51, 177.4]]}, {\"category\": \"car\", \"corners_3d\": [[674.1, 191.19], [667.33, 190.62], [721.85, 190.33], [730.21, 190.87], [674.1, 172.6], [667.33, 172.61], [721.85, 172.61], [730.21, 172.6]]}, {\"category\": \"truck\", \"corners_3d\": [[597.0, 192.22], [597.49, 191.52], [673.03, 191.52], [675.38, 192.22], [597.0, 152.14], [597.49, 152.89], [673.03, 152.89], [675.38, 152.13]]}]\n```" - } - ] - }, - { - "id": 477, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[80.31, 365.72], [-99.28, 365.76], [136.02, 301.67], [255.93, 301.65], [80.31, 203.1], [-99.28, 203.1], [136.02, 193.05], [255.93, 193.05]]}, {\"category\": \"car\", \"corners_3d\": [[901.34, 386.49], [1110.08, 381.42], [2274.13, 816.67], [1687.34, 867.54], [901.34, 189.95], [1110.08, 189.55], [2274.13, 224.39], [1687.34, 228.46]]}, {\"category\": \"car\", \"corners_3d\": [[290.42, 287.3], [181.25, 287.49], [287.94, 258.4], [369.32, 258.3], [290.42, 188.14], [181.25, 188.16], [287.94, 184.28], [369.32, 184.26]]}, {\"category\": \"car\", \"corners_3d\": [[767.72, 278.89], [872.83, 278.28], [1012.59, 328.81], [857.86, 330.15], [767.72, 180.12], [872.83, 180.08], [1012.59, 183.54], [857.86, 183.64]]}, {\"category\": \"car\", \"corners_3d\": [[527.69, 203.08], [503.12, 203.84], [471.79, 201.81], [495.51, 201.14], [527.69, 177.71], [503.12, 177.83], [471.79, 177.5], [495.51, 177.4]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[689.01, 204.32], [678.94, 204.7], [667.34, 204.13], [677.37, 203.76], [689.01, 170.22], [678.94, 170.19], [667.34, 170.24], [677.37, 170.27]]}, {\"category\": \"car\", \"corners_3d\": [[674.1, 191.19], [667.33, 190.62], [721.85, 190.33], [730.21, 190.87], [674.1, 172.6], [667.33, 172.61], [721.85, 172.61], [730.21, 172.6]]}]\n```" - } - ] - }, - { - "id": 478, - "data_source": "Mono3DRefer", - "scene_name": "004752", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[80.31, 365.72], [-99.28, 365.76], [136.02, 301.67], [255.93, 301.65], [80.31, 203.1], [-99.28, 203.1], [136.02, 193.05], [255.93, 193.05]]}, {\"category\": \"car\", \"corners_3d\": [[901.34, 386.49], [1110.08, 381.42], [2274.13, 816.67], [1687.34, 867.54], [901.34, 189.95], [1110.08, 189.55], [2274.13, 224.39], [1687.34, 228.46]]}, {\"category\": \"car\", \"corners_3d\": [[290.42, 287.3], [181.25, 287.49], [287.94, 258.4], [369.32, 258.3], [290.42, 188.14], [181.25, 188.16], [287.94, 184.28], [369.32, 184.26]]}, {\"category\": \"car\", \"corners_3d\": [[767.72, 278.89], [872.83, 278.28], [1012.59, 328.81], [857.86, 330.15], [767.72, 180.12], [872.83, 180.08], [1012.59, 183.54], [857.86, 183.64]]}, {\"category\": \"car\", \"corners_3d\": [[527.69, 203.08], [503.12, 203.84], [471.79, 201.81], [495.51, 201.14], [527.69, 177.71], [503.12, 177.83], [471.79, 177.5], [495.51, 177.4]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[689.01, 204.32], [678.94, 204.7], [667.34, 204.13], [677.37, 203.76], [689.01, 170.22], [678.94, 170.19], [667.34, 170.24], [677.37, 170.27]]}, {\"category\": \"car\", \"corners_3d\": [[674.1, 191.19], [667.33, 190.62], [721.85, 190.33], [730.21, 190.87], [674.1, 172.6], [667.33, 172.61], [721.85, 172.61], [730.21, 172.6]]}, {\"category\": \"truck\", \"corners_3d\": [[597.0, 192.22], [597.49, 191.52], [673.03, 191.52], [675.38, 192.22], [597.0, 152.14], [597.49, 152.89], [673.03, 152.89], [675.38, 152.13]]}]\n```" - } - ] - }, - { - "id": 479, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[598.99, 217.33], [607.36, 216.92], [612.89, 217.35], [604.49, 217.76], [598.99, 181.57], [607.36, 181.49], [612.89, 181.58], [604.49, 181.66]]}]\n```" - } - ] - }, - { - "id": 480, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-640.54, 921.23], [-1279.55, 934.53], [32.6, 401.25], [219.62, 400.03], [-640.54, 331.36], [-1279.55, 334.18], [32.6, 221.23], [219.62, 220.97]]}, {\"category\": \"car\", \"corners_3d\": [[396.86, 310.33], [285.67, 310.98], [364.71, 274.79], [446.59, 274.44], [396.86, 219.85], [285.67, 220.07], [364.71, 207.7], [446.59, 207.58]]}, {\"category\": \"car\", \"corners_3d\": [[472.15, 263.11], [397.98, 263.3], [436.36, 245.64], [496.01, 245.52], [472.15, 190.27], [397.98, 190.3], [436.36, 186.9], [496.01, 186.87]]}, {\"category\": \"car\", \"corners_3d\": [[607.9, 213.36], [639.15, 213.31], [644.15, 217.3], [609.82, 217.36], [607.9, 183.49], [639.15, 183.47], [644.15, 184.52], [609.82, 184.54]]}]\n```" - } - ] - }, - { - "id": 481, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[555.47, 213.36], [517.44, 213.19], [539.57, 207.21], [572.02, 207.34], [555.47, 155.1], [517.44, 155.18], [539.57, 157.79], [572.02, 157.74]]}]\n```" - } - ] - }, - { - "id": 482, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[598.99, 217.33], [607.36, 216.92], [612.89, 217.35], [604.49, 217.76], [598.99, 181.57], [607.36, 181.49], [612.89, 181.58], [604.49, 181.66]]}, {\"category\": \"car\", \"corners_3d\": [[-640.54, 921.23], [-1279.55, 934.53], [32.6, 401.25], [219.62, 400.03], [-640.54, 331.36], [-1279.55, 334.18], [32.6, 221.23], [219.62, 220.97]]}, {\"category\": \"car\", \"corners_3d\": [[396.86, 310.33], [285.67, 310.98], [364.71, 274.79], [446.59, 274.44], [396.86, 219.85], [285.67, 220.07], [364.71, 207.7], [446.59, 207.58]]}, {\"category\": \"car\", \"corners_3d\": [[472.15, 263.11], [397.98, 263.3], [436.36, 245.64], [496.01, 245.52], [472.15, 190.27], [397.98, 190.3], [436.36, 186.9], [496.01, 186.87]]}, {\"category\": \"car\", \"corners_3d\": [[607.9, 213.36], [639.15, 213.31], [644.15, 217.3], [609.82, 217.36], [607.9, 183.49], [639.15, 183.47], [644.15, 184.52], [609.82, 184.54]]}]\n```" - } - ] - }, - { - "id": 483, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[598.99, 217.33], [607.36, 216.92], [612.89, 217.35], [604.49, 217.76], [598.99, 181.57], [607.36, 181.49], [612.89, 181.58], [604.49, 181.66]]}, {\"category\": \"truck\", \"corners_3d\": [[555.47, 213.36], [517.44, 213.19], [539.57, 207.21], [572.02, 207.34], [555.47, 155.1], [517.44, 155.18], [539.57, 157.79], [572.02, 157.74]]}]\n```" - } - ] - }, - { - "id": 484, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-640.54, 921.23], [-1279.55, 934.53], [32.6, 401.25], [219.62, 400.03], [-640.54, 331.36], [-1279.55, 334.18], [32.6, 221.23], [219.62, 220.97]]}, {\"category\": \"car\", \"corners_3d\": [[396.86, 310.33], [285.67, 310.98], [364.71, 274.79], [446.59, 274.44], [396.86, 219.85], [285.67, 220.07], [364.71, 207.7], [446.59, 207.58]]}, {\"category\": \"car\", \"corners_3d\": [[472.15, 263.11], [397.98, 263.3], [436.36, 245.64], [496.01, 245.52], [472.15, 190.27], [397.98, 190.3], [436.36, 186.9], [496.01, 186.87]]}, {\"category\": \"truck\", \"corners_3d\": [[555.47, 213.36], [517.44, 213.19], [539.57, 207.21], [572.02, 207.34], [555.47, 155.1], [517.44, 155.18], [539.57, 157.79], [572.02, 157.74]]}, {\"category\": \"car\", \"corners_3d\": [[607.9, 213.36], [639.15, 213.31], [644.15, 217.3], [609.82, 217.36], [607.9, 183.49], [639.15, 183.47], [644.15, 184.52], [609.82, 184.54]]}]\n```" - } - ] - }, - { - "id": 485, - "data_source": "Mono3DRefer", - "scene_name": "004817", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[598.99, 217.33], [607.36, 216.92], [612.89, 217.35], [604.49, 217.76], [598.99, 181.57], [607.36, 181.49], [612.89, 181.58], [604.49, 181.66]]}, {\"category\": \"car\", \"corners_3d\": [[-640.54, 921.23], [-1279.55, 934.53], [32.6, 401.25], [219.62, 400.03], [-640.54, 331.36], [-1279.55, 334.18], [32.6, 221.23], [219.62, 220.97]]}, {\"category\": \"car\", \"corners_3d\": [[396.86, 310.33], [285.67, 310.98], [364.71, 274.79], [446.59, 274.44], [396.86, 219.85], [285.67, 220.07], [364.71, 207.7], [446.59, 207.58]]}, {\"category\": \"car\", \"corners_3d\": [[472.15, 263.11], [397.98, 263.3], [436.36, 245.64], [496.01, 245.52], [472.15, 190.27], [397.98, 190.3], [436.36, 186.9], [496.01, 186.87]]}, {\"category\": \"truck\", \"corners_3d\": [[555.47, 213.36], [517.44, 213.19], [539.57, 207.21], [572.02, 207.34], [555.47, 155.1], [517.44, 155.18], [539.57, 157.79], [572.02, 157.74]]}, {\"category\": \"car\", \"corners_3d\": [[607.9, 213.36], [639.15, 213.31], [644.15, 217.3], [609.82, 217.36], [607.9, 183.49], [639.15, 183.47], [644.15, 184.52], [609.82, 184.54]]}]\n```" - } - ] - }, - { - "id": 486, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[726.63, 243.97], [736.18, 241.9], [813.47, 246.0], [805.81, 248.34], [726.63, 153.72], [736.18, 154.28], [813.47, 153.18], [805.81, 152.55]]}, {\"category\": \"cyclist\", \"corners_3d\": [[741.01, 259.73], [758.28, 257.29], [854.55, 266.88], [838.36, 269.92], [741.01, 149.83], [758.28, 150.48], [854.55, 147.93], [838.36, 147.13]]}]\n```" - } - ] - }, - { - "id": 487, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[749.26, 256.55], [744.98, 253.97], [767.85, 253.97], [772.86, 256.56], [749.26, 162.22], [744.98, 162.55], [767.85, 162.55], [772.86, 162.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[720.89, 246.68], [707.45, 244.96], [720.73, 244.02], [734.3, 245.7], [720.89, 169.76], [707.45, 169.83], [720.73, 169.87], [734.3, 169.8]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[295.96, 267.9], [286.42, 270.93], [268.09, 270.98], [278.2, 267.94], [295.96, 174.01], [286.42, 174.04], [268.09, 174.04], [278.2, 174.01]]}]\n```" - } - ] - }, - { - "id": 488, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[597.62, 198.68], [622.54, 198.66], [624.72, 200.76], [597.77, 200.78], [597.62, 174.04], [622.54, 174.04], [624.72, 174.14], [597.77, 174.14]]}, {\"category\": \"car\", \"corners_3d\": [[286.49, 334.67], [429.38, 334.69], [168.86, 567.96], [-179.8, 567.81], [286.49, 182.28], [429.38, 182.29], [168.86, 195.88], [-179.8, 195.87]]}]\n```" - } - ] - }, - { - "id": 489, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[726.63, 243.97], [736.18, 241.9], [813.47, 246.0], [805.81, 248.34], [726.63, 153.72], [736.18, 154.28], [813.47, 153.18], [805.81, 152.55]]}, {\"category\": \"cyclist\", \"corners_3d\": [[741.01, 259.73], [758.28, 257.29], [854.55, 266.88], [838.36, 269.92], [741.01, 149.83], [758.28, 150.48], [854.55, 147.93], [838.36, 147.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[749.26, 256.55], [744.98, 253.97], [767.85, 253.97], [772.86, 256.56], [749.26, 162.22], [744.98, 162.55], [767.85, 162.55], [772.86, 162.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[720.89, 246.68], [707.45, 244.96], [720.73, 244.02], [734.3, 245.7], [720.89, 169.76], [707.45, 169.83], [720.73, 169.87], [734.3, 169.8]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[295.96, 267.9], [286.42, 270.93], [268.09, 270.98], [278.2, 267.94], [295.96, 174.01], [286.42, 174.04], [268.09, 174.04], [278.2, 174.01]]}]\n```" - } - ] - }, - { - "id": 490, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[726.63, 243.97], [736.18, 241.9], [813.47, 246.0], [805.81, 248.34], [726.63, 153.72], [736.18, 154.28], [813.47, 153.18], [805.81, 152.55]]}, {\"category\": \"car\", \"corners_3d\": [[597.62, 198.68], [622.54, 198.66], [624.72, 200.76], [597.77, 200.78], [597.62, 174.04], [622.54, 174.04], [624.72, 174.14], [597.77, 174.14]]}, {\"category\": \"car\", \"corners_3d\": [[286.49, 334.67], [429.38, 334.69], [168.86, 567.96], [-179.8, 567.81], [286.49, 182.28], [429.38, 182.29], [168.86, 195.88], [-179.8, 195.87]]}, {\"category\": \"cyclist\", \"corners_3d\": [[741.01, 259.73], [758.28, 257.29], [854.55, 266.88], [838.36, 269.92], [741.01, 149.83], [758.28, 150.48], [854.55, 147.93], [838.36, 147.13]]}]\n```" - } - ] - }, - { - "id": 491, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[597.62, 198.68], [622.54, 198.66], [624.72, 200.76], [597.77, 200.78], [597.62, 174.04], [622.54, 174.04], [624.72, 174.14], [597.77, 174.14]]}, {\"category\": \"car\", \"corners_3d\": [[286.49, 334.67], [429.38, 334.69], [168.86, 567.96], [-179.8, 567.81], [286.49, 182.28], [429.38, 182.29], [168.86, 195.88], [-179.8, 195.87]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[749.26, 256.55], [744.98, 253.97], [767.85, 253.97], [772.86, 256.56], [749.26, 162.22], [744.98, 162.55], [767.85, 162.55], [772.86, 162.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[720.89, 246.68], [707.45, 244.96], [720.73, 244.02], [734.3, 245.7], [720.89, 169.76], [707.45, 169.83], [720.73, 169.87], [734.3, 169.8]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[295.96, 267.9], [286.42, 270.93], [268.09, 270.98], [278.2, 267.94], [295.96, 174.01], [286.42, 174.04], [268.09, 174.04], [278.2, 174.01]]}]\n```" - } - ] - }, - { - "id": 492, - "data_source": "Mono3DRefer", - "scene_name": "004850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[726.63, 243.97], [736.18, 241.9], [813.47, 246.0], [805.81, 248.34], [726.63, 153.72], [736.18, 154.28], [813.47, 153.18], [805.81, 152.55]]}, {\"category\": \"car\", \"corners_3d\": [[597.62, 198.68], [622.54, 198.66], [624.72, 200.76], [597.77, 200.78], [597.62, 174.04], [622.54, 174.04], [624.72, 174.14], [597.77, 174.14]]}, {\"category\": \"car\", \"corners_3d\": [[286.49, 334.67], [429.38, 334.69], [168.86, 567.96], [-179.8, 567.81], [286.49, 182.28], [429.38, 182.29], [168.86, 195.88], [-179.8, 195.87]]}, {\"category\": \"cyclist\", \"corners_3d\": [[741.01, 259.73], [758.28, 257.29], [854.55, 266.88], [838.36, 269.92], [741.01, 149.83], [758.28, 150.48], [854.55, 147.93], [838.36, 147.13]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[749.26, 256.55], [744.98, 253.97], [767.85, 253.97], [772.86, 256.56], [749.26, 162.22], [744.98, 162.55], [767.85, 162.55], [772.86, 162.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[720.89, 246.68], [707.45, 244.96], [720.73, 244.02], [734.3, 245.7], [720.89, 169.76], [707.45, 169.83], [720.73, 169.87], [734.3, 169.8]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[295.96, 267.9], [286.42, 270.93], [268.09, 270.98], [278.2, 267.94], [295.96, 174.01], [286.42, 174.04], [268.09, 174.04], [278.2, 174.01]]}]\n```" - } - ] - }, - { - "id": 493, - "data_source": "Mono3DRefer", - "scene_name": "004861", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[498.47, 244.06], [436.48, 244.19], [465.14, 231.25], [515.87, 231.17], [498.47, 186.93], [436.48, 186.96], [465.14, 184.4], [515.87, 184.38]]}]\n```" - } - ] - }, - { - "id": 494, - "data_source": "Mono3DRefer", - "scene_name": "004881", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[860.53, 284.43], [750.11, 284.79], [705.17, 252.45], [783.87, 252.27], [860.53, 185.38], [750.11, 185.42], [705.17, 181.79], [783.87, 181.77]]}, {\"category\": \"car\", \"corners_3d\": [[453.8, 278.05], [348.46, 278.37], [408.36, 252.68], [487.98, 252.5], [453.8, 187.5], [348.46, 187.54], [408.36, 183.97], [487.98, 183.94]]}, {\"category\": \"car\", \"corners_3d\": [[677.28, 224.8], [733.42, 224.76], [762.14, 236.18], [693.66, 236.24], [677.28, 175.69], [733.42, 175.69], [762.14, 176.31], [693.66, 176.31]]}, {\"category\": \"car\", \"corners_3d\": [[727.07, 220.71], [672.75, 220.71], [663.13, 213.49], [709.24, 213.48], [727.07, 172.5], [672.75, 172.5], [663.13, 172.56], [709.24, 172.56]]}, {\"category\": \"car\", \"corners_3d\": [[765.83, 196.72], [758.92, 195.66], [838.89, 195.66], [849.54, 196.72], [765.83, 164.68], [758.92, 165.05], [838.89, 165.05], [849.54, 164.68]]}]\n```" - } - ] - }, - { - "id": 495, - "data_source": "Mono3DRefer", - "scene_name": "004892", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[767.91, 302.84], [890.57, 301.06], [1176.34, 409.36], [954.79, 415.51], [767.91, 172.07], [890.57, 172.08], [1176.34, 171.42], [954.79, 171.38]]}, {\"category\": \"car\", \"corners_3d\": [[466.86, 302.89], [353.17, 304.53], [407.86, 268.15], [489.94, 267.29], [466.86, 200.55], [353.17, 200.9], [407.86, 193.15], [489.94, 192.97]]}, {\"category\": \"car\", \"corners_3d\": [[499.75, 218.41], [536.15, 218.18], [535.9, 225.26], [493.78, 225.56], [499.75, 183.36], [536.15, 183.31], [535.9, 184.94], [493.78, 185.01]]}, {\"category\": \"car\", \"corners_3d\": [[624.51, 217.55], [663.15, 217.43], [675.21, 223.47], [631.37, 223.62], [624.51, 174.73], [663.15, 174.72], [675.21, 174.98], [631.37, 174.98]]}]\n```" - } - ] - }, - { - "id": 496, - "data_source": "Mono3DRefer", - "scene_name": "004927", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[593.21, 199.15], [621.74, 199.18], [621.47, 201.66], [590.25, 201.63], [593.21, 170.97], [621.74, 170.97], [621.47, 170.79], [590.25, 170.79]]}]\n```" - } - ] - }, - { - "id": 497, - "data_source": "Mono3DRefer", - "scene_name": "004927", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[72.97, 216.5], [65.56, 218.77], [-235.24, 226.18], [-209.57, 223.14], [72.97, 167.16], [65.56, 166.86], [-235.24, 165.9], [-209.57, 166.29]]}]\n```" - } - ] - }, - { - "id": 498, - "data_source": "Mono3DRefer", - "scene_name": "004927", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[593.21, 199.15], [621.74, 199.18], [621.47, 201.66], [590.25, 201.63], [593.21, 170.97], [621.74, 170.97], [621.47, 170.79], [590.25, 170.79]]}, {\"category\": \"truck\", \"corners_3d\": [[72.97, 216.5], [65.56, 218.77], [-235.24, 226.18], [-209.57, 223.14], [72.97, 167.16], [65.56, 166.86], [-235.24, 165.9], [-209.57, 166.29]]}]\n```" - } - ] - }, - { - "id": 499, - "data_source": "Mono3DRefer", - "scene_name": "004932", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[454.25, 325.01], [309.16, 325.34], [410.25, 272.66], [505.14, 272.52], [454.25, 191.74], [309.16, 191.78], [410.25, 185.24], [505.14, 185.22]]}, {\"category\": \"car\", \"corners_3d\": [[682.29, 239.38], [736.42, 239.23], [774.71, 256.41], [706.63, 256.64], [682.29, 182.45], [736.42, 182.43], [774.71, 184.9], [706.63, 184.94]]}, {\"category\": \"car\", \"corners_3d\": [[661.71, 226.43], [705.11, 226.24], [730.11, 235.89], [678.94, 236.16], [661.71, 182.89], [705.11, 182.86], [730.11, 184.66], [678.94, 184.71]]}, {\"category\": \"car\", \"corners_3d\": [[694.02, 195.44], [672.59, 195.48], [665.07, 193.89], [685.01, 193.85], [694.02, 172.98], [672.59, 172.98], [665.07, 172.97], [685.01, 172.97]]}]\n```" - } - ] - }, - { - "id": 500, - "data_source": "Mono3DRefer", - "scene_name": "004932", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[735.36, 270.98], [833.35, 270.46], [974.04, 324.93], [822.09, 326.21], [735.36, 158.17], [833.35, 158.25], [974.04, 150.09], [822.09, 149.9]]}]\n```" - } - ] - }, - { - "id": 501, - "data_source": "Mono3DRefer", - "scene_name": "004932", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[454.25, 325.01], [309.16, 325.34], [410.25, 272.66], [505.14, 272.52], [454.25, 191.74], [309.16, 191.78], [410.25, 185.24], [505.14, 185.22]]}, {\"category\": \"van\", \"corners_3d\": [[735.36, 270.98], [833.35, 270.46], [974.04, 324.93], [822.09, 326.21], [735.36, 158.17], [833.35, 158.25], [974.04, 150.09], [822.09, 149.9]]}, {\"category\": \"car\", \"corners_3d\": [[682.29, 239.38], [736.42, 239.23], [774.71, 256.41], [706.63, 256.64], [682.29, 182.45], [736.42, 182.43], [774.71, 184.9], [706.63, 184.94]]}, {\"category\": \"car\", \"corners_3d\": [[661.71, 226.43], [705.11, 226.24], [730.11, 235.89], [678.94, 236.16], [661.71, 182.89], [705.11, 182.86], [730.11, 184.66], [678.94, 184.71]]}, {\"category\": \"car\", \"corners_3d\": [[694.02, 195.44], [672.59, 195.48], [665.07, 193.89], [685.01, 193.85], [694.02, 172.98], [672.59, 172.98], [665.07, 172.97], [685.01, 172.97]]}]\n```" - } - ] - }, - { - "id": 502, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[644.75, 220.68], [682.57, 220.58], [698.22, 227.96], [654.58, 228.09], [644.75, 180.86], [682.57, 180.84], [698.22, 182.07], [654.58, 182.1]]}, {\"category\": \"van\", \"corners_3d\": [[638.11, 208.45], [665.61, 208.4], [674.98, 212.35], [644.44, 212.42], [638.11, 177.08], [665.61, 177.08], [674.98, 177.55], [644.44, 177.55]]}]\n```" - } - ] - }, - { - "id": 503, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[783.9, 249.48], [799.53, 249.49], [814.44, 255.76], [797.53, 255.74], [783.9, 187.13], [799.53, 187.13], [814.44, 188.3], [797.53, 188.29]]}]\n```" - } - ] - }, - { - "id": 504, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-93.2, 841.18], [-694.26, 847.12], [211.49, 375.9], [391.19, 375.36], [-93.2, 270.36], [-694.26, 271.23], [211.49, 202.48], [391.19, 202.4]]}, {\"category\": \"car\", \"corners_3d\": [[1366.6, 523.51], [1065.83, 528.16], [828.44, 351.89], [982.56, 350.7], [1366.6, 199.92], [1065.83, 200.28], [828.44, 186.67], [982.56, 186.58]]}, {\"category\": \"car\", \"corners_3d\": [[528.86, 255.48], [462.25, 255.79], [481.18, 242.48], [537.06, 242.26], [528.86, 195.15], [462.25, 195.24], [481.18, 191.64], [537.06, 191.58]]}, {\"category\": \"car\", \"corners_3d\": [[665.45, 237.02], [722.49, 236.97], [746.07, 249.61], [677.8, 249.68], [665.45, 181.91], [722.49, 181.91], [746.07, 183.69], [677.8, 183.7]]}, {\"category\": \"car\", \"corners_3d\": [[569.79, 215.12], [540.54, 215.13], [546.44, 211.06], [572.87, 211.05], [569.79, 188.15], [540.54, 188.16], [546.44, 186.69], [572.87, 186.68]]}]\n```" - } - ] - }, - { - "id": 505, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[783.9, 249.48], [799.53, 249.49], [814.44, 255.76], [797.53, 255.74], [783.9, 187.13], [799.53, 187.13], [814.44, 188.3], [797.53, 188.29]]}, {\"category\": \"van\", \"corners_3d\": [[644.75, 220.68], [682.57, 220.58], [698.22, 227.96], [654.58, 228.09], [644.75, 180.86], [682.57, 180.84], [698.22, 182.07], [654.58, 182.1]]}, {\"category\": \"van\", \"corners_3d\": [[638.11, 208.45], [665.61, 208.4], [674.98, 212.35], [644.44, 212.42], [638.11, 177.08], [665.61, 177.08], [674.98, 177.55], [644.44, 177.55]]}]\n```" - } - ] - }, - { - "id": 506, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-93.2, 841.18], [-694.26, 847.12], [211.49, 375.9], [391.19, 375.36], [-93.2, 270.36], [-694.26, 271.23], [211.49, 202.48], [391.19, 202.4]]}, {\"category\": \"car\", \"corners_3d\": [[1366.6, 523.51], [1065.83, 528.16], [828.44, 351.89], [982.56, 350.7], [1366.6, 199.92], [1065.83, 200.28], [828.44, 186.67], [982.56, 186.58]]}, {\"category\": \"car\", \"corners_3d\": [[528.86, 255.48], [462.25, 255.79], [481.18, 242.48], [537.06, 242.26], [528.86, 195.15], [462.25, 195.24], [481.18, 191.64], [537.06, 191.58]]}, {\"category\": \"car\", \"corners_3d\": [[665.45, 237.02], [722.49, 236.97], [746.07, 249.61], [677.8, 249.68], [665.45, 181.91], [722.49, 181.91], [746.07, 183.69], [677.8, 183.7]]}, {\"category\": \"van\", \"corners_3d\": [[644.75, 220.68], [682.57, 220.58], [698.22, 227.96], [654.58, 228.09], [644.75, 180.86], [682.57, 180.84], [698.22, 182.07], [654.58, 182.1]]}, {\"category\": \"car\", \"corners_3d\": [[569.79, 215.12], [540.54, 215.13], [546.44, 211.06], [572.87, 211.05], [569.79, 188.15], [540.54, 188.16], [546.44, 186.69], [572.87, 186.68]]}, {\"category\": \"van\", \"corners_3d\": [[638.11, 208.45], [665.61, 208.4], [674.98, 212.35], [644.44, 212.42], [638.11, 177.08], [665.61, 177.08], [674.98, 177.55], [644.44, 177.55]]}]\n```" - } - ] - }, - { - "id": 507, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-93.2, 841.18], [-694.26, 847.12], [211.49, 375.9], [391.19, 375.36], [-93.2, 270.36], [-694.26, 271.23], [211.49, 202.48], [391.19, 202.4]]}, {\"category\": \"car\", \"corners_3d\": [[1366.6, 523.51], [1065.83, 528.16], [828.44, 351.89], [982.56, 350.7], [1366.6, 199.92], [1065.83, 200.28], [828.44, 186.67], [982.56, 186.58]]}, {\"category\": \"cyclist\", \"corners_3d\": [[783.9, 249.48], [799.53, 249.49], [814.44, 255.76], [797.53, 255.74], [783.9, 187.13], [799.53, 187.13], [814.44, 188.3], [797.53, 188.29]]}, {\"category\": \"car\", \"corners_3d\": [[528.86, 255.48], [462.25, 255.79], [481.18, 242.48], [537.06, 242.26], [528.86, 195.15], [462.25, 195.24], [481.18, 191.64], [537.06, 191.58]]}, {\"category\": \"car\", \"corners_3d\": [[665.45, 237.02], [722.49, 236.97], [746.07, 249.61], [677.8, 249.68], [665.45, 181.91], [722.49, 181.91], [746.07, 183.69], [677.8, 183.7]]}, {\"category\": \"car\", \"corners_3d\": [[569.79, 215.12], [540.54, 215.13], [546.44, 211.06], [572.87, 211.05], [569.79, 188.15], [540.54, 188.16], [546.44, 186.69], [572.87, 186.68]]}]\n```" - } - ] - }, - { - "id": 508, - "data_source": "Mono3DRefer", - "scene_name": "005054", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-93.2, 841.18], [-694.26, 847.12], [211.49, 375.9], [391.19, 375.36], [-93.2, 270.36], [-694.26, 271.23], [211.49, 202.48], [391.19, 202.4]]}, {\"category\": \"car\", \"corners_3d\": [[1366.6, 523.51], [1065.83, 528.16], [828.44, 351.89], [982.56, 350.7], [1366.6, 199.92], [1065.83, 200.28], [828.44, 186.67], [982.56, 186.58]]}, {\"category\": \"cyclist\", \"corners_3d\": [[783.9, 249.48], [799.53, 249.49], [814.44, 255.76], [797.53, 255.74], [783.9, 187.13], [799.53, 187.13], [814.44, 188.3], [797.53, 188.29]]}, {\"category\": \"car\", \"corners_3d\": [[528.86, 255.48], [462.25, 255.79], [481.18, 242.48], [537.06, 242.26], [528.86, 195.15], [462.25, 195.24], [481.18, 191.64], [537.06, 191.58]]}, {\"category\": \"car\", \"corners_3d\": [[665.45, 237.02], [722.49, 236.97], [746.07, 249.61], [677.8, 249.68], [665.45, 181.91], [722.49, 181.91], [746.07, 183.69], [677.8, 183.7]]}, {\"category\": \"van\", \"corners_3d\": [[644.75, 220.68], [682.57, 220.58], [698.22, 227.96], [654.58, 228.09], [644.75, 180.86], [682.57, 180.84], [698.22, 182.07], [654.58, 182.1]]}, {\"category\": \"car\", \"corners_3d\": [[569.79, 215.12], [540.54, 215.13], [546.44, 211.06], [572.87, 211.05], [569.79, 188.15], [540.54, 188.16], [546.44, 186.69], [572.87, 186.68]]}, {\"category\": \"van\", \"corners_3d\": [[638.11, 208.45], [665.61, 208.4], [674.98, 212.35], [644.44, 212.42], [638.11, 177.08], [665.61, 177.08], [674.98, 177.55], [644.44, 177.55]]}]\n```" - } - ] - }, - { - "id": 509, - "data_source": "Mono3DRefer", - "scene_name": "005108", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[316.66, 184.86], [336.36, 184.86], [350.06, 184.84], [329.6, 184.84], [316.66, 168.52], [336.36, 168.79], [350.06, 167.97], [329.6, 167.67]]}, {\"category\": \"car\", \"corners_3d\": [[473.94, 200.12], [507.32, 199.9], [522.67, 201.44], [485.97, 201.72], [473.94, 169.5], [507.32, 169.74], [522.67, 168.11], [485.97, 167.82]]}, {\"category\": \"car\", \"corners_3d\": [[399.43, 195.74], [423.68, 195.54], [443.49, 196.23], [417.99, 196.46], [399.43, 173.54], [423.68, 173.76], [443.49, 172.99], [417.99, 172.74]]}]\n```" - } - ] - }, - { - "id": 510, - "data_source": "Mono3DRefer", - "scene_name": "005120", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[826.72, 278.27], [948.41, 279.68], [1196.39, 374.14], [963.66, 369.21], [826.72, 156.61], [948.41, 156.39], [1196.39, 141.83], [963.66, 142.59]]}, {\"category\": \"car\", \"corners_3d\": [[788.01, 222.41], [844.04, 223.11], [867.19, 235.35], [797.36, 234.25], [788.01, 166.99], [844.04, 166.91], [867.19, 165.46], [797.36, 165.59]]}]\n```" - } - ] - }, - { - "id": 511, - "data_source": "Mono3DRefer", - "scene_name": "005120", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[800.48, 200.45], [844.0, 201.15], [826.13, 205.92], [776.0, 204.97], [800.48, 145.88], [844.0, 145.2], [826.13, 140.53], [776.0, 141.46]]}]\n```" - } - ] - }, - { - "id": 512, - "data_source": "Mono3DRefer", - "scene_name": "005120", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[826.72, 278.27], [948.41, 279.68], [1196.39, 374.14], [963.66, 369.21], [826.72, 156.61], [948.41, 156.39], [1196.39, 141.83], [963.66, 142.59]]}, {\"category\": \"car\", \"corners_3d\": [[788.01, 222.41], [844.04, 223.11], [867.19, 235.35], [797.36, 234.25], [788.01, 166.99], [844.04, 166.91], [867.19, 165.46], [797.36, 165.59]]}, {\"category\": \"van\", \"corners_3d\": [[800.48, 200.45], [844.0, 201.15], [826.13, 205.92], [776.0, 204.97], [800.48, 145.88], [844.0, 145.2], [826.13, 140.53], [776.0, 141.46]]}]\n```" - } - ] - }, - { - "id": 513, - "data_source": "Mono3DRefer", - "scene_name": "005145", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[674.13, 225.74], [727.35, 225.74], [750.39, 236.14], [686.71, 236.14], [674.13, 176.67], [727.35, 176.67], [750.39, 177.42], [686.71, 177.42]]}, {\"category\": \"car\", \"corners_3d\": [[724.25, 270.55], [825.85, 270.42], [950.56, 325.0], [792.26, 325.31], [724.25, 165.68], [825.85, 165.69], [950.56, 161.69], [792.26, 161.66]]}, {\"category\": \"car\", \"corners_3d\": [[237.06, 325.92], [141.2, 367.75], [-398.16, 372.03], [-182.91, 328.55], [237.06, 188.22], [141.2, 192.42], [-398.16, 192.85], [-182.91, 188.49]]}, {\"category\": \"car\", \"corners_3d\": [[-67.19, 313.56], [43.98, 291.87], [304.34, 294.65], [238.84, 317.47], [-67.19, 191.64], [43.98, 188.75], [304.34, 189.12], [238.84, 192.17]]}, {\"category\": \"car\", \"corners_3d\": [[85.46, 277.02], [152.58, 264.24], [344.81, 265.02], [304.19, 278.04], [85.46, 181.14], [152.58, 180.12], [344.81, 180.19], [304.19, 181.22]]}, {\"category\": \"car\", \"corners_3d\": [[440.05, 243.72], [419.92, 251.18], [211.67, 250.5], [251.33, 243.16], [440.05, 161.93], [419.92, 160.78], [211.67, 160.89], [251.33, 162.02]]}, {\"category\": \"car\", \"corners_3d\": [[186.35, 265.16], [231.74, 255.45], [390.16, 255.66], [363.31, 265.43], [186.35, 176.06], [231.74, 175.72], [390.16, 175.73], [363.31, 176.06]]}, {\"category\": \"car\", \"corners_3d\": [[694.21, 241.64], [764.31, 241.31], [809.33, 257.5], [722.81, 258.01], [694.21, 180.1], [764.31, 180.07], [809.33, 181.78], [722.81, 181.83]]}, {\"category\": \"car\", \"corners_3d\": [[273.1, 239.7], [300.75, 234.19], [431.25, 234.17], [415.34, 239.68], [273.1, 164.38], [300.75, 165.08], [431.25, 165.08], [415.34, 164.38]]}, {\"category\": \"car\", \"corners_3d\": [[328.4, 227.25], [349.38, 223.59], [476.03, 223.96], [464.12, 227.69], [328.4, 172.84], [349.38, 172.84], [476.03, 172.84], [464.12, 172.84]]}, {\"category\": \"car\", \"corners_3d\": [[509.39, 210.68], [501.75, 212.6], [406.31, 212.23], [418.45, 210.34], [509.39, 175.11], [501.75, 175.23], [406.31, 175.2], [418.45, 175.09]]}, {\"category\": \"car\", \"corners_3d\": [[438.05, 205.16], [445.49, 203.77], [521.21, 203.77], [517.18, 205.17], [438.05, 166.51], [445.49, 166.78], [521.21, 166.78], [517.18, 166.51]]}, {\"category\": \"car\", \"corners_3d\": [[648.18, 206.66], [681.3, 206.63], [691.2, 210.54], [654.26, 210.57], [648.18, 176.39], [681.3, 176.39], [691.2, 176.8], [654.26, 176.8]]}]\n```" - } - ] - }, - { - "id": 514, - "data_source": "Mono3DRefer", - "scene_name": "005206", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[510.16, 245.32], [438.21, 245.47], [470.0, 230.7], [527.29, 230.61], [510.16, 182.01], [438.21, 182.03], [470.0, 180.16], [527.29, 180.15]]}, {\"category\": \"car\", \"corners_3d\": [[715.09, 223.71], [672.76, 223.69], [665.63, 217.28], [702.62, 217.3], [715.09, 180.34], [672.76, 180.34], [665.63, 179.39], [702.62, 179.4]]}, {\"category\": \"car\", \"corners_3d\": [[563.96, 214.11], [528.25, 213.97], [543.55, 209.18], [575.14, 209.29], [563.96, 182.79], [528.25, 182.75], [543.55, 181.6], [575.14, 181.62]]}]\n```" - } - ] - }, - { - "id": 515, - "data_source": "Mono3DRefer", - "scene_name": "005349", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[860.55, 288.13], [975.6, 306.45], [775.09, 342.89], [667.69, 314.28], [860.55, 167.55], [975.6, 166.71], [775.09, 165.03], [667.69, 166.35]]}, {\"category\": \"car\", \"corners_3d\": [[1033.7, 295.76], [921.06, 279.99], [1059.12, 262.28], [1167.46, 273.01], [1033.7, 164.46], [921.06, 165.54], [1059.12, 166.75], [1167.46, 166.01]]}, {\"category\": \"car\", \"corners_3d\": [[1234.55, 247.12], [1337.25, 254.27], [1267.19, 266.97], [1157.09, 257.55], [1234.55, 178.87], [1337.25, 179.45], [1267.19, 180.48], [1157.09, 179.72]]}]\n```" - } - ] - }, - { - "id": 516, - "data_source": "Mono3DRefer", - "scene_name": "005365", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[178.67, 231.67], [225.76, 229.4], [251.78, 234.78], [201.17, 237.5], [178.67, 188.47], [225.76, 187.87], [251.78, 189.3], [201.17, 190.02]]}, {\"category\": \"car\", \"corners_3d\": [[148.26, 228.58], [192.88, 226.67], [212.72, 231.94], [164.33, 234.25], [148.26, 190.51], [192.88, 189.91], [212.72, 191.58], [164.33, 192.31]]}, {\"category\": \"car\", \"corners_3d\": [[118.07, 216.92], [152.65, 215.9], [157.28, 219.38], [119.96, 220.57], [118.07, 190.22], [152.65, 189.82], [157.28, 191.19], [119.96, 191.66]]}, {\"category\": \"car\", \"corners_3d\": [[215.03, 242.32], [278.89, 238.93], [313.63, 248.49], [242.04, 252.97], [215.03, 187.78], [278.89, 187.05], [313.63, 189.11], [242.04, 190.07]]}]\n```" - } - ] - }, - { - "id": 517, - "data_source": "Mono3DRefer", - "scene_name": "005441", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[489.56, 262.88], [404.79, 262.57], [451.65, 243.95], [518.91, 244.15], [489.56, 187.42], [404.79, 187.37], [451.65, 184.36], [518.91, 184.39]]}, {\"category\": \"car\", \"corners_3d\": [[536.88, 197.33], [560.06, 197.3], [558.5, 199.1], [533.61, 199.13], [536.88, 173.89], [560.06, 173.89], [558.5, 173.97], [533.61, 173.97]]}]\n```" - } - ] - }, - { - "id": 518, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"truck\", \"corners_3d\": [[555.94, 191.66], [557.93, 190.95], [636.63, 190.95], [637.74, 191.66], [555.94, 149.83], [557.93, 150.7], [636.63, 150.7], [637.74, 149.83]]}]\n```" - } - ] - }, - { - "id": 519, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[693.04, 205.88], [681.46, 206.29], [669.59, 205.6], [681.07, 205.21], [693.04, 168.66], [681.46, 168.61], [669.59, 168.7], [681.07, 168.75]]}]\n```" - } - ] - }, - { - "id": 520, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[521.61, 205.22], [491.39, 205.96], [466.72, 203.29], [495.06, 202.66], [521.61, 176.51], [491.39, 176.6], [466.72, 176.29], [495.06, 176.22]]}, {\"category\": \"car\", \"corners_3d\": [[560.19, 193.06], [548.9, 192.5], [598.79, 191.69], [610.97, 192.2], [560.19, 172.14], [548.9, 172.16], [598.79, 172.19], [610.97, 172.17]]}, {\"category\": \"car\", \"corners_3d\": [[418.35, 202.31], [398.35, 203.42], [343.94, 202.16], [365.04, 201.14], [418.35, 178.27], [398.35, 178.47], [343.94, 178.24], [365.04, 178.05]]}, {\"category\": \"car\", \"corners_3d\": [[662.42, 192.57], [654.07, 191.94], [710.07, 191.51], [720.02, 192.11], [662.42, 172.98], [654.07, 172.98], [710.07, 172.97], [720.02, 172.98]]}, {\"category\": \"car\", \"corners_3d\": [[-325.31, 510.93], [-643.9, 512.52], [-64.05, 354.41], [105.16, 353.95], [-325.31, 227.51], [-643.9, 227.77], [-64.05, 202.21], [105.16, 202.13]]}, {\"category\": \"car\", \"corners_3d\": [[167.65, 327.08], [16.96, 328.05], [197.4, 278.95], [299.85, 278.5], [167.65, 192.67], [16.96, 192.79], [197.4, 186.48], [299.85, 186.43]]}, {\"category\": \"car\", \"corners_3d\": [[808.63, 313.85], [947.1, 312.5], [1227.39, 417.41], [987.9, 421.59], [808.63, 182.52], [947.1, 182.43], [1227.39, 189.62], [987.9, 189.9]]}]\n```" - } - ] - }, - { - "id": 521, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[693.04, 205.88], [681.46, 206.29], [669.59, 205.6], [681.07, 205.21], [693.04, 168.66], [681.46, 168.61], [669.59, 168.7], [681.07, 168.75]]}, {\"category\": \"truck\", \"corners_3d\": [[555.94, 191.66], [557.93, 190.95], [636.63, 190.95], [637.74, 191.66], [555.94, 149.83], [557.93, 150.7], [636.63, 150.7], [637.74, 149.83]]}]\n```" - } - ] - }, - { - "id": 522, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[521.61, 205.22], [491.39, 205.96], [466.72, 203.29], [495.06, 202.66], [521.61, 176.51], [491.39, 176.6], [466.72, 176.29], [495.06, 176.22]]}, {\"category\": \"car\", \"corners_3d\": [[560.19, 193.06], [548.9, 192.5], [598.79, 191.69], [610.97, 192.2], [560.19, 172.14], [548.9, 172.16], [598.79, 172.19], [610.97, 172.17]]}, {\"category\": \"car\", \"corners_3d\": [[418.35, 202.31], [398.35, 203.42], [343.94, 202.16], [365.04, 201.14], [418.35, 178.27], [398.35, 178.47], [343.94, 178.24], [365.04, 178.05]]}, {\"category\": \"car\", \"corners_3d\": [[662.42, 192.57], [654.07, 191.94], [710.07, 191.51], [720.02, 192.11], [662.42, 172.98], [654.07, 172.98], [710.07, 172.97], [720.02, 172.98]]}, {\"category\": \"truck\", \"corners_3d\": [[555.94, 191.66], [557.93, 190.95], [636.63, 190.95], [637.74, 191.66], [555.94, 149.83], [557.93, 150.7], [636.63, 150.7], [637.74, 149.83]]}, {\"category\": \"car\", \"corners_3d\": [[-325.31, 510.93], [-643.9, 512.52], [-64.05, 354.41], [105.16, 353.95], [-325.31, 227.51], [-643.9, 227.77], [-64.05, 202.21], [105.16, 202.13]]}, {\"category\": \"car\", \"corners_3d\": [[167.65, 327.08], [16.96, 328.05], [197.4, 278.95], [299.85, 278.5], [167.65, 192.67], [16.96, 192.79], [197.4, 186.48], [299.85, 186.43]]}, {\"category\": \"car\", \"corners_3d\": [[808.63, 313.85], [947.1, 312.5], [1227.39, 417.41], [987.9, 421.59], [808.63, 182.52], [947.1, 182.43], [1227.39, 189.62], [987.9, 189.9]]}]\n```" - } - ] - }, - { - "id": 523, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[521.61, 205.22], [491.39, 205.96], [466.72, 203.29], [495.06, 202.66], [521.61, 176.51], [491.39, 176.6], [466.72, 176.29], [495.06, 176.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[693.04, 205.88], [681.46, 206.29], [669.59, 205.6], [681.07, 205.21], [693.04, 168.66], [681.46, 168.61], [669.59, 168.7], [681.07, 168.75]]}, {\"category\": \"car\", \"corners_3d\": [[560.19, 193.06], [548.9, 192.5], [598.79, 191.69], [610.97, 192.2], [560.19, 172.14], [548.9, 172.16], [598.79, 172.19], [610.97, 172.17]]}, {\"category\": \"car\", \"corners_3d\": [[418.35, 202.31], [398.35, 203.42], [343.94, 202.16], [365.04, 201.14], [418.35, 178.27], [398.35, 178.47], [343.94, 178.24], [365.04, 178.05]]}, {\"category\": \"car\", \"corners_3d\": [[662.42, 192.57], [654.07, 191.94], [710.07, 191.51], [720.02, 192.11], [662.42, 172.98], [654.07, 172.98], [710.07, 172.97], [720.02, 172.98]]}, {\"category\": \"car\", \"corners_3d\": [[-325.31, 510.93], [-643.9, 512.52], [-64.05, 354.41], [105.16, 353.95], [-325.31, 227.51], [-643.9, 227.77], [-64.05, 202.21], [105.16, 202.13]]}, {\"category\": \"car\", \"corners_3d\": [[167.65, 327.08], [16.96, 328.05], [197.4, 278.95], [299.85, 278.5], [167.65, 192.67], [16.96, 192.79], [197.4, 186.48], [299.85, 186.43]]}, {\"category\": \"car\", \"corners_3d\": [[808.63, 313.85], [947.1, 312.5], [1227.39, 417.41], [987.9, 421.59], [808.63, 182.52], [947.1, 182.43], [1227.39, 189.62], [987.9, 189.9]]}]\n```" - } - ] - }, - { - "id": 524, - "data_source": "Mono3DRefer", - "scene_name": "005453", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (truck, pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[521.61, 205.22], [491.39, 205.96], [466.72, 203.29], [495.06, 202.66], [521.61, 176.51], [491.39, 176.6], [466.72, 176.29], [495.06, 176.22]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[693.04, 205.88], [681.46, 206.29], [669.59, 205.6], [681.07, 205.21], [693.04, 168.66], [681.46, 168.61], [669.59, 168.7], [681.07, 168.75]]}, {\"category\": \"car\", \"corners_3d\": [[560.19, 193.06], [548.9, 192.5], [598.79, 191.69], [610.97, 192.2], [560.19, 172.14], [548.9, 172.16], [598.79, 172.19], [610.97, 172.17]]}, {\"category\": \"car\", \"corners_3d\": [[418.35, 202.31], [398.35, 203.42], [343.94, 202.16], [365.04, 201.14], [418.35, 178.27], [398.35, 178.47], [343.94, 178.24], [365.04, 178.05]]}, {\"category\": \"car\", \"corners_3d\": [[662.42, 192.57], [654.07, 191.94], [710.07, 191.51], [720.02, 192.11], [662.42, 172.98], [654.07, 172.98], [710.07, 172.97], [720.02, 172.98]]}, {\"category\": \"truck\", \"corners_3d\": [[555.94, 191.66], [557.93, 190.95], [636.63, 190.95], [637.74, 191.66], [555.94, 149.83], [557.93, 150.7], [636.63, 150.7], [637.74, 149.83]]}, {\"category\": \"car\", \"corners_3d\": [[-325.31, 510.93], [-643.9, 512.52], [-64.05, 354.41], [105.16, 353.95], [-325.31, 227.51], [-643.9, 227.77], [-64.05, 202.21], [105.16, 202.13]]}, {\"category\": \"car\", \"corners_3d\": [[167.65, 327.08], [16.96, 328.05], [197.4, 278.95], [299.85, 278.5], [167.65, 192.67], [16.96, 192.79], [197.4, 186.48], [299.85, 186.43]]}, {\"category\": \"car\", \"corners_3d\": [[808.63, 313.85], [947.1, 312.5], [1227.39, 417.41], [987.9, 421.59], [808.63, 182.52], [947.1, 182.43], [1227.39, 189.62], [987.9, 189.9]]}]\n```" - } - ] - }, - { - "id": 525, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[436.44, 267.57], [461.48, 265.67], [472.39, 268.71], [446.75, 270.73], [436.44, 182.0], [461.48, 181.82], [472.39, 182.11], [446.75, 182.31]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[859.37, 395.07], [902.05, 384.86], [951.44, 399.13], [908.29, 410.79], [859.37, 171.48], [902.05, 171.54], [951.44, 171.45], [908.29, 171.38]]}]\n```" - } - ] - }, - { - "id": 526, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[864.55, 342.78], [1040.1, 349.5], [1317.2, 509.42], [974.44, 485.83], [864.55, 188.65], [1040.1, 189.28], [1317.2, 204.14], [974.44, 201.95]]}, {\"category\": \"car\", \"corners_3d\": [[805.67, 218.01], [851.47, 218.56], [871.73, 228.31], [816.02, 227.49], [805.67, 174.83], [851.47, 174.86], [871.73, 175.28], [816.02, 175.25]]}, {\"category\": \"car\", \"corners_3d\": [[344.04, 261.79], [383.56, 254.26], [511.36, 258.65], [482.02, 267.07], [344.04, 185.75], [383.56, 184.66], [511.36, 185.29], [482.02, 186.51]]}]\n```" - } - ] - }, - { - "id": 527, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[790.9, 200.76], [823.38, 201.02], [830.33, 204.31], [794.02, 204.0], [790.9, 164.65], [823.38, 164.57], [830.33, 163.6], [794.02, 163.7]]}]\n```" - } - ] - }, - { - "id": 528, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[436.44, 267.57], [461.48, 265.67], [472.39, 268.71], [446.75, 270.73], [436.44, 182.0], [461.48, 181.82], [472.39, 182.11], [446.75, 182.31]]}, {\"category\": \"car\", \"corners_3d\": [[864.55, 342.78], [1040.1, 349.5], [1317.2, 509.42], [974.44, 485.83], [864.55, 188.65], [1040.1, 189.28], [1317.2, 204.14], [974.44, 201.95]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[859.37, 395.07], [902.05, 384.86], [951.44, 399.13], [908.29, 410.79], [859.37, 171.48], [902.05, 171.54], [951.44, 171.45], [908.29, 171.38]]}, {\"category\": \"car\", \"corners_3d\": [[805.67, 218.01], [851.47, 218.56], [871.73, 228.31], [816.02, 227.49], [805.67, 174.83], [851.47, 174.86], [871.73, 175.28], [816.02, 175.25]]}, {\"category\": \"car\", \"corners_3d\": [[344.04, 261.79], [383.56, 254.26], [511.36, 258.65], [482.02, 267.07], [344.04, 185.75], [383.56, 184.66], [511.36, 185.29], [482.02, 186.51]]}]\n```" - } - ] - }, - { - "id": 529, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[436.44, 267.57], [461.48, 265.67], [472.39, 268.71], [446.75, 270.73], [436.44, 182.0], [461.48, 181.82], [472.39, 182.11], [446.75, 182.31]]}, {\"category\": \"van\", \"corners_3d\": [[790.9, 200.76], [823.38, 201.02], [830.33, 204.31], [794.02, 204.0], [790.9, 164.65], [823.38, 164.57], [830.33, 163.6], [794.02, 163.7]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[859.37, 395.07], [902.05, 384.86], [951.44, 399.13], [908.29, 410.79], [859.37, 171.48], [902.05, 171.54], [951.44, 171.45], [908.29, 171.38]]}]\n```" - } - ] - }, - { - "id": 530, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[790.9, 200.76], [823.38, 201.02], [830.33, 204.31], [794.02, 204.0], [790.9, 164.65], [823.38, 164.57], [830.33, 163.6], [794.02, 163.7]]}, {\"category\": \"car\", \"corners_3d\": [[864.55, 342.78], [1040.1, 349.5], [1317.2, 509.42], [974.44, 485.83], [864.55, 188.65], [1040.1, 189.28], [1317.2, 204.14], [974.44, 201.95]]}, {\"category\": \"car\", \"corners_3d\": [[805.67, 218.01], [851.47, 218.56], [871.73, 228.31], [816.02, 227.49], [805.67, 174.83], [851.47, 174.86], [871.73, 175.28], [816.02, 175.25]]}, {\"category\": \"car\", \"corners_3d\": [[344.04, 261.79], [383.56, 254.26], [511.36, 258.65], [482.02, 267.07], [344.04, 185.75], [383.56, 184.66], [511.36, 185.29], [482.02, 186.51]]}]\n```" - } - ] - }, - { - "id": 531, - "data_source": "Mono3DRefer", - "scene_name": "005515", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[436.44, 267.57], [461.48, 265.67], [472.39, 268.71], [446.75, 270.73], [436.44, 182.0], [461.48, 181.82], [472.39, 182.11], [446.75, 182.31]]}, {\"category\": \"van\", \"corners_3d\": [[790.9, 200.76], [823.38, 201.02], [830.33, 204.31], [794.02, 204.0], [790.9, 164.65], [823.38, 164.57], [830.33, 163.6], [794.02, 163.7]]}, {\"category\": \"car\", \"corners_3d\": [[864.55, 342.78], [1040.1, 349.5], [1317.2, 509.42], [974.44, 485.83], [864.55, 188.65], [1040.1, 189.28], [1317.2, 204.14], [974.44, 201.95]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[859.37, 395.07], [902.05, 384.86], [951.44, 399.13], [908.29, 410.79], [859.37, 171.48], [902.05, 171.54], [951.44, 171.45], [908.29, 171.38]]}, {\"category\": \"car\", \"corners_3d\": [[805.67, 218.01], [851.47, 218.56], [871.73, 228.31], [816.02, 227.49], [805.67, 174.83], [851.47, 174.86], [871.73, 175.28], [816.02, 175.25]]}, {\"category\": \"car\", \"corners_3d\": [[344.04, 261.79], [383.56, 254.26], [511.36, 258.65], [482.02, 267.07], [344.04, 185.75], [383.56, 184.66], [511.36, 185.29], [482.02, 186.51]]}]\n```" - } - ] - }, - { - "id": 532, - "data_source": "Mono3DRefer", - "scene_name": "005540", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[579.4, 318.29], [592.34, 328.71], [433.23, 337.9], [431.39, 326.19], [579.4, 176.97], [592.34, 176.71], [433.23, 176.47], [431.39, 176.77]]}]\n```" - } - ] - }, - { - "id": 533, - "data_source": "Mono3DRefer", - "scene_name": "005540", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[126.07, 270.75], [137.86, 273.12], [85.02, 277.99], [74.04, 275.37], [126.07, 168.51], [137.86, 168.2], [85.02, 167.55], [74.04, 167.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[195.77, 257.68], [205.77, 259.62], [152.24, 263.78], [143.11, 261.62], [195.77, 170.25], [205.77, 169.99], [152.24, 169.44], [143.11, 169.72]]}]\n```" - } - ] - }, - { - "id": 534, - "data_source": "Mono3DRefer", - "scene_name": "005540", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[579.4, 318.29], [592.34, 328.71], [433.23, 337.9], [431.39, 326.19], [579.4, 176.97], [592.34, 176.71], [433.23, 176.47], [431.39, 176.77]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[126.07, 270.75], [137.86, 273.12], [85.02, 277.99], [74.04, 275.37], [126.07, 168.51], [137.86, 168.2], [85.02, 167.55], [74.04, 167.9]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[195.77, 257.68], [205.77, 259.62], [152.24, 263.78], [143.11, 261.62], [195.77, 170.25], [205.77, 169.99], [152.24, 169.44], [143.11, 169.72]]}]\n```" - } - ] - }, - { - "id": 535, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[586.22, 198.51], [610.42, 198.49], [611.59, 200.52], [585.47, 200.54], [586.22, 174.59], [610.42, 174.59], [611.59, 174.72], [585.47, 174.73]]}]\n```" - } - ] - }, - { - "id": 536, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[696.86, 213.75], [686.94, 213.79], [679.41, 211.39], [688.76, 211.35], [696.86, 163.63], [686.94, 163.63], [679.41, 164.17], [688.76, 164.17]]}]\n```" - } - ] - }, - { - "id": 537, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[121.84, 234.73], [43.06, 235.12], [345.5, 199.09], [377.99, 199.02], [121.84, 134.2], [43.06, 133.95], [345.5, 156.47], [377.99, 156.51]]}]\n```" - } - ] - }, - { - "id": 538, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[586.22, 198.51], [610.42, 198.49], [611.59, 200.52], [585.47, 200.54], [586.22, 174.59], [610.42, 174.59], [611.59, 174.72], [585.47, 174.73]]}, {\"category\": \"cyclist\", \"corners_3d\": [[696.86, 213.75], [686.94, 213.79], [679.41, 211.39], [688.76, 211.35], [696.86, 163.63], [686.94, 163.63], [679.41, 164.17], [688.76, 164.17]]}]\n```" - } - ] - }, - { - "id": 539, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[586.22, 198.51], [610.42, 198.49], [611.59, 200.52], [585.47, 200.54], [586.22, 174.59], [610.42, 174.59], [611.59, 174.72], [585.47, 174.73]]}, {\"category\": \"tram\", \"corners_3d\": [[121.84, 234.73], [43.06, 235.12], [345.5, 199.09], [377.99, 199.02], [121.84, 134.2], [43.06, 133.95], [345.5, 156.47], [377.99, 156.51]]}]\n```" - } - ] - }, - { - "id": 540, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"tram\", \"corners_3d\": [[121.84, 234.73], [43.06, 235.12], [345.5, 199.09], [377.99, 199.02], [121.84, 134.2], [43.06, 133.95], [345.5, 156.47], [377.99, 156.51]]}, {\"category\": \"cyclist\", \"corners_3d\": [[696.86, 213.75], [686.94, 213.79], [679.41, 211.39], [688.76, 211.35], [696.86, 163.63], [686.94, 163.63], [679.41, 164.17], [688.76, 164.17]]}]\n```" - } - ] - }, - { - "id": 541, - "data_source": "Mono3DRefer", - "scene_name": "005555", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist, tram). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[586.22, 198.51], [610.42, 198.49], [611.59, 200.52], [585.47, 200.54], [586.22, 174.59], [610.42, 174.59], [611.59, 174.72], [585.47, 174.73]]}, {\"category\": \"tram\", \"corners_3d\": [[121.84, 234.73], [43.06, 235.12], [345.5, 199.09], [377.99, 199.02], [121.84, 134.2], [43.06, 133.95], [345.5, 156.47], [377.99, 156.51]]}, {\"category\": \"cyclist\", \"corners_3d\": [[696.86, 213.75], [686.94, 213.79], [679.41, 211.39], [688.76, 211.35], [696.86, 163.63], [686.94, 163.63], [679.41, 164.17], [688.76, 164.17]]}]\n```" - } - ] - }, - { - "id": 542, - "data_source": "Mono3DRefer", - "scene_name": "005616", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[475.62, 234.16], [423.6, 234.55], [440.62, 225.82], [485.23, 225.53], [475.62, 192.07], [423.6, 192.19], [440.62, 189.45], [485.23, 189.36]]}, {\"category\": \"car\", \"corners_3d\": [[495.42, 219.78], [452.49, 220.01], [464.54, 214.24], [502.2, 214.07], [495.42, 177.97], [452.49, 178.0], [464.54, 177.37], [502.2, 177.35]]}, {\"category\": \"car\", \"corners_3d\": [[856.08, 261.52], [758.63, 262.91], [698.54, 239.31], [771.43, 238.55], [856.08, 157.08], [758.63, 156.83], [698.54, 161.03], [771.43, 161.17]]}, {\"category\": \"car\", \"corners_3d\": [[408.83, 275.32], [323.92, 276.28], [378.27, 251.54], [442.63, 250.99], [408.83, 194.99], [323.92, 195.2], [378.27, 189.86], [442.63, 189.74]]}]\n```" - } - ] - }, - { - "id": 543, - "data_source": "Mono3DRefer", - "scene_name": "005745", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[580.78, 216.22], [545.44, 215.97], [560.25, 211.73], [592.17, 211.93], [580.78, 184.48], [545.44, 184.42], [560.25, 183.28], [592.17, 183.33]]}]\n```" - } - ] - }, - { - "id": 544, - "data_source": "Mono3DRefer", - "scene_name": "005747", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[772.47, 206.71], [782.3, 208.59], [676.78, 208.7], [672.48, 206.81], [772.47, 172.37], [782.3, 172.35], [676.78, 172.34], [672.48, 172.37]]}, {\"category\": \"car\", \"corners_3d\": [[648.33, 193.34], [648.28, 192.67], [703.12, 192.75], [705.02, 193.42], [648.33, 172.42], [648.28, 172.44], [703.12, 172.44], [705.02, 172.42]]}]\n```" - } - ] - }, - { - "id": 545, - "data_source": "Mono3DRefer", - "scene_name": "005801", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[852.95, 288.1], [869.77, 295.37], [780.68, 295.37], [769.5, 288.1], [852.95, 154.58], [869.77, 152.83], [780.68, 152.83], [769.5, 154.58]]}]\n```" - } - ] - }, - { - "id": 546, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[678.76, 237.87], [742.26, 237.7], [786.88, 256.51], [705.04, 256.79], [678.76, 170.92], [742.26, 170.92], [786.88, 170.36], [705.04, 170.35]]}, {\"category\": \"van\", \"corners_3d\": [[657.1, 211.7], [694.94, 211.64], [711.93, 217.82], [668.08, 217.91], [657.1, 168.53], [694.94, 168.54], [711.93, 167.85], [668.08, 167.84]]}]\n```" - } - ] - }, - { - "id": 547, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[922.84, 290.56], [948.57, 290.52], [998.09, 307.38], [968.69, 307.43], [922.84, 187.19], [948.57, 187.18], [998.09, 189.23], [968.69, 189.24]]}]\n```" - } - ] - }, - { - "id": 548, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.35, 316.73], [911.16, 316.76], [1127.22, 420.04], [878.44, 419.97], [766.35, 177.08], [911.16, 177.08], [1127.22, 180.12], [878.44, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[713.54, 264.13], [796.88, 264.03], [864.92, 296.01], [752.39, 296.19], [713.54, 188.33], [796.88, 188.31], [864.92, 193.73], [752.39, 193.76]]}, {\"category\": \"car\", \"corners_3d\": [[563.87, 217.13], [521.79, 217.15], [532.42, 211.26], [568.91, 211.24], [563.87, 178.35], [521.79, 178.35], [532.42, 177.62], [568.91, 177.62]]}, {\"category\": \"car\", \"corners_3d\": [[346.47, 421.87], [108.71, 423.57], [305.47, 322.03], [446.53, 321.43], [346.47, 207.96], [108.71, 208.2], [305.47, 193.89], [446.53, 193.8]]}]\n```" - } - ] - }, - { - "id": 549, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[678.76, 237.87], [742.26, 237.7], [786.88, 256.51], [705.04, 256.79], [678.76, 170.92], [742.26, 170.92], [786.88, 170.36], [705.04, 170.35]]}, {\"category\": \"van\", \"corners_3d\": [[657.1, 211.7], [694.94, 211.64], [711.93, 217.82], [668.08, 217.91], [657.1, 168.53], [694.94, 168.54], [711.93, 167.85], [668.08, 167.84]]}, {\"category\": \"cyclist\", \"corners_3d\": [[922.84, 290.56], [948.57, 290.52], [998.09, 307.38], [968.69, 307.43], [922.84, 187.19], [948.57, 187.18], [998.09, 189.23], [968.69, 189.24]]}]\n```" - } - ] - }, - { - "id": 550, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.35, 316.73], [911.16, 316.76], [1127.22, 420.04], [878.44, 419.97], [766.35, 177.08], [911.16, 177.08], [1127.22, 180.12], [878.44, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[713.54, 264.13], [796.88, 264.03], [864.92, 296.01], [752.39, 296.19], [713.54, 188.33], [796.88, 188.31], [864.92, 193.73], [752.39, 193.76]]}, {\"category\": \"van\", \"corners_3d\": [[678.76, 237.87], [742.26, 237.7], [786.88, 256.51], [705.04, 256.79], [678.76, 170.92], [742.26, 170.92], [786.88, 170.36], [705.04, 170.35]]}, {\"category\": \"car\", \"corners_3d\": [[563.87, 217.13], [521.79, 217.15], [532.42, 211.26], [568.91, 211.24], [563.87, 178.35], [521.79, 178.35], [532.42, 177.62], [568.91, 177.62]]}, {\"category\": \"van\", \"corners_3d\": [[657.1, 211.7], [694.94, 211.64], [711.93, 217.82], [668.08, 217.91], [657.1, 168.53], [694.94, 168.54], [711.93, 167.85], [668.08, 167.84]]}, {\"category\": \"car\", \"corners_3d\": [[346.47, 421.87], [108.71, 423.57], [305.47, 322.03], [446.53, 321.43], [346.47, 207.96], [108.71, 208.2], [305.47, 193.89], [446.53, 193.8]]}]\n```" - } - ] - }, - { - "id": 551, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.35, 316.73], [911.16, 316.76], [1127.22, 420.04], [878.44, 419.97], [766.35, 177.08], [911.16, 177.08], [1127.22, 180.12], [878.44, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[713.54, 264.13], [796.88, 264.03], [864.92, 296.01], [752.39, 296.19], [713.54, 188.33], [796.88, 188.31], [864.92, 193.73], [752.39, 193.76]]}, {\"category\": \"car\", \"corners_3d\": [[563.87, 217.13], [521.79, 217.15], [532.42, 211.26], [568.91, 211.24], [563.87, 178.35], [521.79, 178.35], [532.42, 177.62], [568.91, 177.62]]}, {\"category\": \"cyclist\", \"corners_3d\": [[922.84, 290.56], [948.57, 290.52], [998.09, 307.38], [968.69, 307.43], [922.84, 187.19], [948.57, 187.18], [998.09, 189.23], [968.69, 189.24]]}, {\"category\": \"car\", \"corners_3d\": [[346.47, 421.87], [108.71, 423.57], [305.47, 322.03], [446.53, 321.43], [346.47, 207.96], [108.71, 208.2], [305.47, 193.89], [446.53, 193.8]]}]\n```" - } - ] - }, - { - "id": 552, - "data_source": "Mono3DRefer", - "scene_name": "005822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[766.35, 316.73], [911.16, 316.76], [1127.22, 420.04], [878.44, 419.97], [766.35, 177.08], [911.16, 177.08], [1127.22, 180.12], [878.44, 180.11]]}, {\"category\": \"car\", \"corners_3d\": [[713.54, 264.13], [796.88, 264.03], [864.92, 296.01], [752.39, 296.19], [713.54, 188.33], [796.88, 188.31], [864.92, 193.73], [752.39, 193.76]]}, {\"category\": \"van\", \"corners_3d\": [[678.76, 237.87], [742.26, 237.7], [786.88, 256.51], [705.04, 256.79], [678.76, 170.92], [742.26, 170.92], [786.88, 170.36], [705.04, 170.35]]}, {\"category\": \"car\", \"corners_3d\": [[563.87, 217.13], [521.79, 217.15], [532.42, 211.26], [568.91, 211.24], [563.87, 178.35], [521.79, 178.35], [532.42, 177.62], [568.91, 177.62]]}, {\"category\": \"van\", \"corners_3d\": [[657.1, 211.7], [694.94, 211.64], [711.93, 217.82], [668.08, 217.91], [657.1, 168.53], [694.94, 168.54], [711.93, 167.85], [668.08, 167.84]]}, {\"category\": \"cyclist\", \"corners_3d\": [[922.84, 290.56], [948.57, 290.52], [998.09, 307.38], [968.69, 307.43], [922.84, 187.19], [948.57, 187.18], [998.09, 189.23], [968.69, 189.24]]}, {\"category\": \"car\", \"corners_3d\": [[346.47, 421.87], [108.71, 423.57], [305.47, 322.03], [446.53, 321.43], [346.47, 207.96], [108.71, 208.2], [305.47, 193.89], [446.53, 193.8]]}]\n```" - } - ] - }, - { - "id": 553, - "data_source": "Mono3DRefer", - "scene_name": "005869", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[543.54, 238.68], [498.4, 238.76], [511.77, 229.6], [550.62, 229.54], [543.54, 197.01], [498.4, 197.04], [511.77, 193.68], [550.62, 193.66]]}]\n```" - } - ] - }, - { - "id": 554, - "data_source": "Mono3DRefer", - "scene_name": "005887", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[541.42, 228.64], [490.02, 228.37], [513.28, 220.75], [557.7, 220.96], [541.42, 182.65], [490.02, 182.61], [513.28, 181.27], [557.7, 181.3]]}]\n```" - } - ] - }, - { - "id": 555, - "data_source": "Mono3DRefer", - "scene_name": "005907", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[486.72, 209.49], [452.36, 209.61], [462.97, 205.93], [493.87, 205.83], [486.72, 179.13], [452.36, 179.15], [462.97, 178.52], [493.87, 178.5]]}]\n```" - } - ] - }, - { - "id": 556, - "data_source": "Mono3DRefer", - "scene_name": "005913", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[750.14, 215.73], [794.21, 216.24], [791.27, 220.74], [740.99, 220.07], [750.14, 176.95], [794.21, 176.81], [791.27, 175.59], [740.99, 175.77]]}, {\"category\": \"car\", \"corners_3d\": [[12.96, 249.13], [65.16, 247.77], [22.72, 257.83], [-39.15, 259.67], [12.96, 213.63], [65.16, 213.03], [22.72, 217.5], [-39.15, 218.32]]}, {\"category\": \"car\", \"corners_3d\": [[95.71, 237.08], [138.55, 236.1], [122.58, 240.77], [75.39, 241.94], [95.71, 201.8], [138.55, 201.48], [122.58, 202.98], [75.39, 203.35]]}, {\"category\": \"car\", \"corners_3d\": [[802.7, 196.5], [824.25, 196.63], [818.71, 197.35], [795.89, 197.21], [802.7, 178.02], [824.25, 177.94], [818.71, 177.48], [795.89, 177.57]]}, {\"category\": \"car\", \"corners_3d\": [[867.01, 187.32], [883.99, 187.34], [879.36, 187.44], [861.64, 187.42], [867.01, 173.77], [883.99, 173.65], [879.36, 173.11], [861.64, 173.24]]}]\n```" - } - ] - }, - { - "id": 557, - "data_source": "Mono3DRefer", - "scene_name": "005969", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[652.97, 219.84], [688.28, 219.82], [701.62, 227.15], [660.79, 227.18], [652.97, 188.29], [688.28, 188.28], [701.62, 190.69], [660.79, 190.7]]}]\n```" - } - ] - }, - { - "id": 558, - "data_source": "Mono3DRefer", - "scene_name": "005975", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[563.61, 207.87], [533.83, 207.8], [543.15, 204.87], [570.44, 204.93], [563.61, 177.73], [533.83, 177.72], [543.15, 177.32], [570.44, 177.33]]}, {\"category\": \"van\", \"corners_3d\": [[629.61, 204.01], [654.19, 204.05], [655.48, 206.71], [628.81, 206.66], [629.61, 177.36], [654.19, 177.37], [655.48, 177.76], [628.81, 177.75]]}]\n```" - } - ] - }, - { - "id": 559, - "data_source": "Mono3DRefer", - "scene_name": "005975", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-971.94, 1417.9], [-2183.91, 1419.57], [71.59, 412.74], [304.46, 412.68], [-971.94, 157.76], [-2183.91, 157.74], [71.59, 169.95], [304.46, 169.95]]}, {\"category\": \"car\", \"corners_3d\": [[394.15, 272.44], [488.31, 272.71], [431.98, 313.99], [299.26, 313.45], [394.15, 193.21], [488.31, 193.26], [431.98, 201.7], [299.26, 201.59]]}, {\"category\": \"car\", \"corners_3d\": [[684.97, 229.26], [640.13, 229.4], [632.53, 222.24], [671.72, 222.13], [684.97, 187.64], [640.13, 187.68], [632.53, 185.8], [671.72, 185.78]]}]\n```" - } - ] - }, - { - "id": 560, - "data_source": "Mono3DRefer", - "scene_name": "005975", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-971.94, 1417.9], [-2183.91, 1419.57], [71.59, 412.74], [304.46, 412.68], [-971.94, 157.76], [-2183.91, 157.74], [71.59, 169.95], [304.46, 169.95]]}, {\"category\": \"car\", \"corners_3d\": [[394.15, 272.44], [488.31, 272.71], [431.98, 313.99], [299.26, 313.45], [394.15, 193.21], [488.31, 193.26], [431.98, 201.7], [299.26, 201.59]]}, {\"category\": \"car\", \"corners_3d\": [[684.97, 229.26], [640.13, 229.4], [632.53, 222.24], [671.72, 222.13], [684.97, 187.64], [640.13, 187.68], [632.53, 185.8], [671.72, 185.78]]}, {\"category\": \"van\", \"corners_3d\": [[563.61, 207.87], [533.83, 207.8], [543.15, 204.87], [570.44, 204.93], [563.61, 177.73], [533.83, 177.72], [543.15, 177.32], [570.44, 177.33]]}, {\"category\": \"van\", \"corners_3d\": [[629.61, 204.01], [654.19, 204.05], [655.48, 206.71], [628.81, 206.66], [629.61, 177.36], [654.19, 177.37], [655.48, 177.76], [628.81, 177.75]]}]\n```" - } - ] - }, - { - "id": 561, - "data_source": "Mono3DRefer", - "scene_name": "005978", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[764.08, 289.09], [879.94, 288.92], [1101.13, 381.6], [893.1, 382.16], [764.08, 185.34], [879.94, 185.32], [1101.13, 195.27], [893.1, 195.33]]}]\n```" - } - ] - }, - { - "id": 562, - "data_source": "Mono3DRefer", - "scene_name": "006013", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[305.95, 363.48], [127.05, 363.51], [308.66, 291.66], [420.14, 291.65], [305.95, 168.51], [127.05, 168.51], [308.66, 170.14], [420.14, 170.14]]}]\n```" - } - ] - }, - { - "id": 563, - "data_source": "Mono3DRefer", - "scene_name": "006013", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-1183.29, 1308.36], [-2287.67, 1372.77], [-52.89, 437.71], [173.88, 434.43], [-1183.29, 416.45], [-2287.67, 430.27], [-52.89, 229.67], [173.88, 228.97]]}, {\"category\": \"car\", \"corners_3d\": [[537.08, 230.43], [483.18, 230.44], [502.42, 221.63], [548.08, 221.63], [537.08, 185.47], [483.18, 185.47], [502.42, 183.54], [548.08, 183.54]]}]\n```" - } - ] - }, - { - "id": 564, - "data_source": "Mono3DRefer", - "scene_name": "006013", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-1183.29, 1308.36], [-2287.67, 1372.77], [-52.89, 437.71], [173.88, 434.43], [-1183.29, 416.45], [-2287.67, 430.27], [-52.89, 229.67], [173.88, 228.97]]}, {\"category\": \"van\", \"corners_3d\": [[305.95, 363.48], [127.05, 363.51], [308.66, 291.66], [420.14, 291.65], [305.95, 168.51], [127.05, 168.51], [308.66, 170.14], [420.14, 170.14]]}, {\"category\": \"car\", \"corners_3d\": [[537.08, 230.43], [483.18, 230.44], [502.42, 221.63], [548.08, 221.63], [537.08, 185.47], [483.18, 185.47], [502.42, 183.54], [548.08, 183.54]]}]\n```" - } - ] - }, - { - "id": 565, - "data_source": "Mono3DRefer", - "scene_name": "006058", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[220.32, 234.55], [249.91, 229.96], [364.02, 230.06], [343.57, 234.67], [220.32, 175.06], [249.91, 174.89], [364.02, 174.9], [343.57, 175.06]]}, {\"category\": \"car\", \"corners_3d\": [[741.55, 202.39], [703.29, 202.52], [685.06, 198.96], [718.82, 198.86], [741.55, 162.02], [703.29, 161.97], [685.06, 163.28], [718.82, 163.31]]}, {\"category\": \"car\", \"corners_3d\": [[563.73, 205.81], [531.26, 205.9], [535.01, 202.3], [563.94, 202.23], [563.73, 174.36], [531.26, 174.37], [535.01, 174.2], [563.94, 174.2]]}]\n```" - } - ] - }, - { - "id": 566, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[588.61, 188.66], [628.56, 188.66], [635.12, 190.01], [587.63, 190.01], [588.61, 138.99], [628.56, 139.01], [635.12, 131.0], [587.63, 130.97]]}]\n```" - } - ] - }, - { - "id": 567, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[473.91, 252.62], [427.97, 252.39], [452.82, 243.92], [493.32, 244.09], [473.91, 158.43], [427.97, 158.5], [452.82, 161.26], [493.32, 161.2]]}, {\"category\": \"cyclist\", \"corners_3d\": [[386.8, 273.35], [341.0, 273.41], [380.78, 258.88], [419.32, 258.84], [386.8, 154.45], [341.0, 154.43], [380.78, 158.71], [419.32, 158.72]]}]\n```" - } - ] - }, - { - "id": 568, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[757.96, 260.43], [814.57, 260.12], [836.17, 266.92], [774.73, 267.29], [757.96, 158.66], [814.57, 158.75], [836.17, 156.78], [774.73, 156.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[671.87, 212.9], [703.63, 212.89], [706.41, 213.68], [673.84, 213.7], [671.87, 153.27], [703.63, 153.28], [706.41, 152.56], [673.84, 152.55]]}]\n```" - } - ] - }, - { - "id": 569, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[473.91, 252.62], [427.97, 252.39], [452.82, 243.92], [493.32, 244.09], [473.91, 158.43], [427.97, 158.5], [452.82, 161.26], [493.32, 161.2]]}, {\"category\": \"cyclist\", \"corners_3d\": [[386.8, 273.35], [341.0, 273.41], [380.78, 258.88], [419.32, 258.84], [386.8, 154.45], [341.0, 154.43], [380.78, 158.71], [419.32, 158.72]]}, {\"category\": \"van\", \"corners_3d\": [[588.61, 188.66], [628.56, 188.66], [635.12, 190.01], [587.63, 190.01], [588.61, 138.99], [628.56, 139.01], [635.12, 131.0], [587.63, 130.97]]}]\n```" - } - ] - }, - { - "id": 570, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[757.96, 260.43], [814.57, 260.12], [836.17, 266.92], [774.73, 267.29], [757.96, 158.66], [814.57, 158.75], [836.17, 156.78], [774.73, 156.68]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[671.87, 212.9], [703.63, 212.89], [706.41, 213.68], [673.84, 213.7], [671.87, 153.27], [703.63, 153.28], [706.41, 152.56], [673.84, 152.55]]}, {\"category\": \"van\", \"corners_3d\": [[588.61, 188.66], [628.56, 188.66], [635.12, 190.01], [587.63, 190.01], [588.61, 138.99], [628.56, 139.01], [635.12, 131.0], [587.63, 130.97]]}]\n```" - } - ] - }, - { - "id": 571, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[757.96, 260.43], [814.57, 260.12], [836.17, 266.92], [774.73, 267.29], [757.96, 158.66], [814.57, 158.75], [836.17, 156.78], [774.73, 156.68]]}, {\"category\": \"cyclist\", \"corners_3d\": [[473.91, 252.62], [427.97, 252.39], [452.82, 243.92], [493.32, 244.09], [473.91, 158.43], [427.97, 158.5], [452.82, 161.26], [493.32, 161.2]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[671.87, 212.9], [703.63, 212.89], [706.41, 213.68], [673.84, 213.7], [671.87, 153.27], [703.63, 153.28], [706.41, 152.56], [673.84, 152.55]]}, {\"category\": \"cyclist\", \"corners_3d\": [[386.8, 273.35], [341.0, 273.41], [380.78, 258.88], [419.32, 258.84], [386.8, 154.45], [341.0, 154.43], [380.78, 158.71], [419.32, 158.72]]}]\n```" - } - ] - }, - { - "id": 572, - "data_source": "Mono3DRefer", - "scene_name": "006071", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[757.96, 260.43], [814.57, 260.12], [836.17, 266.92], [774.73, 267.29], [757.96, 158.66], [814.57, 158.75], [836.17, 156.78], [774.73, 156.68]]}, {\"category\": \"cyclist\", \"corners_3d\": [[473.91, 252.62], [427.97, 252.39], [452.82, 243.92], [493.32, 244.09], [473.91, 158.43], [427.97, 158.5], [452.82, 161.26], [493.32, 161.2]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[671.87, 212.9], [703.63, 212.89], [706.41, 213.68], [673.84, 213.7], [671.87, 153.27], [703.63, 153.28], [706.41, 152.56], [673.84, 152.55]]}, {\"category\": \"cyclist\", \"corners_3d\": [[386.8, 273.35], [341.0, 273.41], [380.78, 258.88], [419.32, 258.84], [386.8, 154.45], [341.0, 154.43], [380.78, 158.71], [419.32, 158.72]]}, {\"category\": \"van\", \"corners_3d\": [[588.61, 188.66], [628.56, 188.66], [635.12, 190.01], [587.63, 190.01], [588.61, 138.99], [628.56, 139.01], [635.12, 131.0], [587.63, 130.97]]}]\n```" - } - ] - }, - { - "id": 573, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[475.26, 233.4], [439.5, 233.45], [453.3, 228.54], [485.68, 228.49], [475.26, 161.34], [439.5, 161.32], [453.3, 163.23], [485.68, 163.25]]}, {\"category\": \"cyclist\", \"corners_3d\": [[414.05, 246.37], [380.7, 246.49], [403.45, 238.73], [432.78, 238.63], [414.05, 160.46], [380.7, 160.42], [403.45, 162.94], [432.78, 162.97]]}]\n```" - } - ] - }, - { - "id": 574, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[552.38, 192.19], [591.71, 192.15], [599.27, 194.11], [552.71, 194.16], [552.38, 143.24], [591.71, 143.39], [599.27, 136.38], [552.71, 136.18]]}]\n```" - } - ] - }, - { - "id": 575, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[743.1, 272.64], [802.58, 271.87], [827.83, 280.21], [763.05, 281.12], [743.1, 163.84], [802.58, 163.99], [827.83, 162.37], [763.05, 162.2]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[646.1, 220.34], [678.12, 220.24], [681.19, 221.23], [648.36, 221.34], [646.1, 159.81], [678.12, 159.87], [681.19, 159.32], [648.36, 159.26]]}]\n```" - } - ] - }, - { - "id": 576, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[475.26, 233.4], [439.5, 233.45], [453.3, 228.54], [485.68, 228.49], [475.26, 161.34], [439.5, 161.32], [453.3, 163.23], [485.68, 163.25]]}, {\"category\": \"cyclist\", \"corners_3d\": [[414.05, 246.37], [380.7, 246.49], [403.45, 238.73], [432.78, 238.63], [414.05, 160.46], [380.7, 160.42], [403.45, 162.94], [432.78, 162.97]]}, {\"category\": \"van\", \"corners_3d\": [[552.38, 192.19], [591.71, 192.15], [599.27, 194.11], [552.71, 194.16], [552.38, 143.24], [591.71, 143.39], [599.27, 136.38], [552.71, 136.18]]}]\n```" - } - ] - }, - { - "id": 577, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[743.1, 272.64], [802.58, 271.87], [827.83, 280.21], [763.05, 281.12], [743.1, 163.84], [802.58, 163.99], [827.83, 162.37], [763.05, 162.2]]}, {\"category\": \"cyclist\", \"corners_3d\": [[475.26, 233.4], [439.5, 233.45], [453.3, 228.54], [485.68, 228.49], [475.26, 161.34], [439.5, 161.32], [453.3, 163.23], [485.68, 163.25]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[646.1, 220.34], [678.12, 220.24], [681.19, 221.23], [648.36, 221.34], [646.1, 159.81], [678.12, 159.87], [681.19, 159.32], [648.36, 159.26]]}, {\"category\": \"cyclist\", \"corners_3d\": [[414.05, 246.37], [380.7, 246.49], [403.45, 238.73], [432.78, 238.63], [414.05, 160.46], [380.7, 160.42], [403.45, 162.94], [432.78, 162.97]]}]\n```" - } - ] - }, - { - "id": 578, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[743.1, 272.64], [802.58, 271.87], [827.83, 280.21], [763.05, 281.12], [743.1, 163.84], [802.58, 163.99], [827.83, 162.37], [763.05, 162.2]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[646.1, 220.34], [678.12, 220.24], [681.19, 221.23], [648.36, 221.34], [646.1, 159.81], [678.12, 159.87], [681.19, 159.32], [648.36, 159.26]]}, {\"category\": \"van\", \"corners_3d\": [[552.38, 192.19], [591.71, 192.15], [599.27, 194.11], [552.71, 194.16], [552.38, 143.24], [591.71, 143.39], [599.27, 136.38], [552.71, 136.18]]}]\n```" - } - ] - }, - { - "id": 579, - "data_source": "Mono3DRefer", - "scene_name": "006083", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, van, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[743.1, 272.64], [802.58, 271.87], [827.83, 280.21], [763.05, 281.12], [743.1, 163.84], [802.58, 163.99], [827.83, 162.37], [763.05, 162.2]]}, {\"category\": \"cyclist\", \"corners_3d\": [[475.26, 233.4], [439.5, 233.45], [453.3, 228.54], [485.68, 228.49], [475.26, 161.34], [439.5, 161.32], [453.3, 163.23], [485.68, 163.25]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[646.1, 220.34], [678.12, 220.24], [681.19, 221.23], [648.36, 221.34], [646.1, 159.81], [678.12, 159.87], [681.19, 159.32], [648.36, 159.26]]}, {\"category\": \"cyclist\", \"corners_3d\": [[414.05, 246.37], [380.7, 246.49], [403.45, 238.73], [432.78, 238.63], [414.05, 160.46], [380.7, 160.42], [403.45, 162.94], [432.78, 162.97]]}, {\"category\": \"van\", \"corners_3d\": [[552.38, 192.19], [591.71, 192.15], [599.27, 194.11], [552.71, 194.16], [552.38, 143.24], [591.71, 143.39], [599.27, 136.38], [552.71, 136.18]]}]\n```" - } - ] - }, - { - "id": 580, - "data_source": "Mono3DRefer", - "scene_name": "006195", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[786.97, 324.41], [928.58, 321.69], [1363.12, 490.3], [1071.98, 502.94], [786.97, 171.94], [928.58, 171.95], [1363.12, 170.93], [1071.98, 170.85]]}, {\"category\": \"car\", \"corners_3d\": [[436.97, 322.67], [305.31, 324.84], [380.5, 278.35], [471.54, 277.3], [436.97, 204.76], [305.31, 205.22], [380.5, 195.32], [471.54, 195.1]]}, {\"category\": \"car\", \"corners_3d\": [[488.61, 219.27], [526.73, 219.0], [526.13, 226.57], [481.71, 226.93], [488.61, 182.62], [526.73, 182.56], [526.13, 184.16], [481.71, 184.23]]}, {\"category\": \"car\", \"corners_3d\": [[621.77, 220.19], [662.0, 220.03], [675.49, 226.72], [629.59, 226.92], [621.77, 175.54], [662.0, 175.53], [675.49, 175.91], [629.59, 175.93]]}]\n```" - } - ] - }, - { - "id": 581, - "data_source": "Mono3DRefer", - "scene_name": "006244", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[88.04, 219.76], [48.95, 220.96], [52.43, 217.21], [88.47, 216.19], [88.04, 192.11], [48.95, 192.6], [52.43, 191.06], [88.47, 190.64]]}]\n```" - } - ] - }, - { - "id": 582, - "data_source": "Mono3DRefer", - "scene_name": "006266", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[333.83, 348.28], [152.76, 348.31], [324.6, 282.23], [437.46, 282.21], [333.83, 184.95], [152.76, 184.96], [324.6, 180.4], [437.46, 180.4]]}, {\"category\": \"car\", \"corners_3d\": [[738.94, 282.39], [850.57, 282.4], [987.11, 344.61], [812.08, 344.57], [738.94, 190.52], [850.57, 190.52], [987.11, 200.56], [812.08, 200.55]]}, {\"category\": \"car\", \"corners_3d\": [[754.6, 239.91], [687.69, 240.04], [669.34, 226.8], [723.11, 226.72], [754.6, 180.01], [687.69, 180.02], [669.34, 178.61], [723.11, 178.6]]}, {\"category\": \"car\", \"corners_3d\": [[507.72, 234.1], [441.05, 234.33], [464.55, 223.94], [519.92, 223.78], [507.72, 176.87], [441.05, 176.88], [464.55, 176.2], [519.92, 176.19]]}, {\"category\": \"car\", \"corners_3d\": [[659.0, 215.77], [701.34, 215.75], [716.83, 222.5], [667.84, 222.53], [659.0, 178.75], [701.34, 178.74], [716.83, 179.67], [667.84, 179.67]]}, {\"category\": \"car\", \"corners_3d\": [[697.26, 211.26], [655.8, 211.3], [648.46, 206.7], [684.97, 206.66], [697.26, 174.4], [655.8, 174.4], [648.46, 174.22], [684.97, 174.22]]}]\n```" - } - ] - }, - { - "id": 583, - "data_source": "Mono3DRefer", - "scene_name": "006269", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[99.77, 597.93], [-366.81, 623.87], [196.33, 342.72], [365.94, 338.91], [99.77, 228.44], [-366.81, 231.83], [196.33, 195.07], [365.94, 194.57]]}, {\"category\": \"car\", \"corners_3d\": [[474.17, 252.65], [380.15, 252.34], [440.92, 233.06], [512.24, 233.23], [474.17, 163.97], [380.15, 164.0], [440.92, 166.15], [512.24, 166.13]]}, {\"category\": \"car\", \"corners_3d\": [[663.19, 197.75], [692.66, 198.12], [679.71, 199.92], [648.36, 199.51], [663.19, 167.44], [692.66, 167.36], [679.71, 166.97], [648.36, 167.06]]}]\n```" - } - ] - }, - { - "id": 584, - "data_source": "Mono3DRefer", - "scene_name": "006269", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[407.1, 302.47], [262.65, 304.52], [358.93, 260.58], [454.67, 259.67], [407.1, 150.67], [262.65, 150.32], [358.93, 157.84], [454.67, 158.0]]}]\n```" - } - ] - }, - { - "id": 585, - "data_source": "Mono3DRefer", - "scene_name": "006269", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[99.77, 597.93], [-366.81, 623.87], [196.33, 342.72], [365.94, 338.91], [99.77, 228.44], [-366.81, 231.83], [196.33, 195.07], [365.94, 194.57]]}, {\"category\": \"van\", \"corners_3d\": [[407.1, 302.47], [262.65, 304.52], [358.93, 260.58], [454.67, 259.67], [407.1, 150.67], [262.65, 150.32], [358.93, 157.84], [454.67, 158.0]]}, {\"category\": \"car\", \"corners_3d\": [[474.17, 252.65], [380.15, 252.34], [440.92, 233.06], [512.24, 233.23], [474.17, 163.97], [380.15, 164.0], [440.92, 166.15], [512.24, 166.13]]}, {\"category\": \"car\", \"corners_3d\": [[663.19, 197.75], [692.66, 198.12], [679.71, 199.92], [648.36, 199.51], [663.19, 167.44], [692.66, 167.36], [679.71, 166.97], [648.36, 167.06]]}]\n```" - } - ] - }, - { - "id": 586, - "data_source": "Mono3DRefer", - "scene_name": "006275", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[428.64, 278.55], [321.2, 279.49], [388.06, 250.33], [465.87, 249.83], [428.64, 185.65], [321.2, 185.76], [388.06, 182.23], [465.87, 182.17]]}, {\"category\": \"car\", \"corners_3d\": [[702.3, 227.54], [651.02, 227.7], [640.42, 219.5], [684.07, 219.38], [702.3, 174.8], [651.02, 174.8], [640.42, 174.51], [684.07, 174.51]]}, {\"category\": \"car\", \"corners_3d\": [[524.06, 214.99], [481.82, 214.94], [500.89, 209.27], [537.46, 209.31], [524.06, 178.18], [481.82, 178.17], [500.89, 177.45], [537.46, 177.46]]}, {\"category\": \"car\", \"corners_3d\": [[622.9, 210.76], [657.42, 210.54], [673.88, 215.16], [635.2, 215.43], [622.9, 178.36], [657.42, 178.32], [673.88, 179.0], [635.2, 179.04]]}]\n```" - } - ] - }, - { - "id": 587, - "data_source": "Mono3DRefer", - "scene_name": "006332", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[751.19, 228.32], [771.6, 230.11], [715.59, 233.7], [695.79, 231.69], [751.19, 204.42], [771.6, 205.44], [715.59, 207.48], [695.79, 206.34]]}]\n```" - } - ] - }, - { - "id": 588, - "data_source": "Mono3DRefer", - "scene_name": "006332", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[623.65, 226.35], [656.47, 226.23], [666.69, 232.84], [629.81, 232.99], [623.65, 189.18], [656.47, 189.15], [666.69, 191.17], [629.81, 191.21]]}, {\"category\": \"van\", \"corners_3d\": [[551.33, 211.04], [570.42, 210.93], [572.99, 213.38], [552.68, 213.5], [551.33, 187.47], [570.42, 187.43], [572.99, 188.36], [552.68, 188.41]]}]\n```" - } - ] - }, - { - "id": 589, - "data_source": "Mono3DRefer", - "scene_name": "006332", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[623.65, 226.35], [656.47, 226.23], [666.69, 232.84], [629.81, 232.99], [623.65, 189.18], [656.47, 189.15], [666.69, 191.17], [629.81, 191.21]]}, {\"category\": \"car\", \"corners_3d\": [[751.19, 228.32], [771.6, 230.11], [715.59, 233.7], [695.79, 231.69], [751.19, 204.42], [771.6, 205.44], [715.59, 207.48], [695.79, 206.34]]}, {\"category\": \"van\", \"corners_3d\": [[551.33, 211.04], [570.42, 210.93], [572.99, 213.38], [552.68, 213.5], [551.33, 187.47], [570.42, 187.43], [572.99, 188.36], [552.68, 188.41]]}]\n```" - } - ] - }, - { - "id": 590, - "data_source": "Mono3DRefer", - "scene_name": "006349", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[660.08, 213.28], [688.98, 213.28], [698.78, 218.3], [666.28, 218.3], [660.08, 187.48], [688.98, 187.48], [698.78, 189.3], [666.28, 189.3]]}]\n```" - } - ] - }, - { - "id": 591, - "data_source": "Mono3DRefer", - "scene_name": "006370", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[690.73, 262.54], [788.14, 261.94], [901.02, 308.9], [752.89, 310.3], [690.73, 172.83], [788.14, 172.83], [901.02, 172.82], [752.89, 172.82]]}, {\"category\": \"car\", \"corners_3d\": [[-18.55, 271.42], [52.85, 260.49], [230.64, 260.96], [181.12, 272.02], [-18.55, 187.65], [52.85, 186.01], [230.64, 186.08], [181.12, 187.74]]}, {\"category\": \"car\", \"corners_3d\": [[487.29, 218.76], [537.42, 218.54], [533.44, 226.77], [474.21, 227.07], [487.29, 173.13], [537.42, 173.13], [533.44, 173.18], [474.21, 173.18]]}, {\"category\": \"car\", \"corners_3d\": [[640.65, 220.14], [688.56, 219.79], [719.56, 229.0], [662.45, 229.49], [640.65, 172.84], [688.56, 172.84], [719.56, 172.84], [662.45, 172.84]]}, {\"category\": \"car\", \"corners_3d\": [[528.6, 202.22], [557.38, 202.31], [548.06, 204.96], [516.72, 204.86], [528.6, 175.08], [557.38, 175.08], [548.06, 175.29], [516.72, 175.28]]}, {\"category\": \"car\", \"corners_3d\": [[622.29, 199.87], [652.65, 199.91], [655.19, 203.21], [621.13, 203.16], [622.29, 168.17], [652.65, 168.16], [655.19, 167.59], [621.13, 167.6]]}]\n```" - } - ] - }, - { - "id": 592, - "data_source": "Mono3DRefer", - "scene_name": "006381", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1084.36, 320.51], [914.85, 323.09], [777.99, 265.41], [884.57, 264.42], [1084.36, 138.19], [914.85, 137.58], [777.99, 151.12], [884.57, 151.36]]}, {\"category\": \"car\", \"corners_3d\": [[331.25, 339.13], [197.19, 340.37], [324.77, 284.56], [413.75, 284.01], [331.25, 211.54], [197.19, 211.83], [324.77, 198.84], [413.75, 198.71]]}, {\"category\": \"car\", \"corners_3d\": [[471.22, 252.99], [403.99, 253.37], [433.82, 239.14], [489.11, 238.89], [471.22, 198.26], [403.99, 198.38], [433.82, 193.87], [489.11, 193.79]]}, {\"category\": \"car\", \"corners_3d\": [[503.69, 231.83], [451.22, 232.01], [470.07, 223.36], [514.84, 223.23], [503.69, 180.41], [451.22, 180.44], [470.07, 179.33], [514.84, 179.31]]}, {\"category\": \"car\", \"corners_3d\": [[596.2, 202.06], [622.41, 202.0], [626.94, 204.37], [598.59, 204.44], [596.2, 176.97], [622.41, 176.96], [626.94, 177.3], [598.59, 177.31]]}]\n```" - } - ] - }, - { - "id": 593, - "data_source": "Mono3DRefer", - "scene_name": "006381", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[591.37, 202.62], [598.35, 202.39], [602.94, 202.63], [595.93, 202.86], [591.37, 172.85], [598.35, 172.85], [602.94, 172.85], [595.93, 172.85]]}]\n```" - } - ] - }, - { - "id": 594, - "data_source": "Mono3DRefer", - "scene_name": "006381", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1084.36, 320.51], [914.85, 323.09], [777.99, 265.41], [884.57, 264.42], [1084.36, 138.19], [914.85, 137.58], [777.99, 151.12], [884.57, 151.36]]}, {\"category\": \"car\", \"corners_3d\": [[331.25, 339.13], [197.19, 340.37], [324.77, 284.56], [413.75, 284.01], [331.25, 211.54], [197.19, 211.83], [324.77, 198.84], [413.75, 198.71]]}, {\"category\": \"car\", \"corners_3d\": [[471.22, 252.99], [403.99, 253.37], [433.82, 239.14], [489.11, 238.89], [471.22, 198.26], [403.99, 198.38], [433.82, 193.87], [489.11, 193.79]]}, {\"category\": \"car\", \"corners_3d\": [[503.69, 231.83], [451.22, 232.01], [470.07, 223.36], [514.84, 223.23], [503.69, 180.41], [451.22, 180.44], [470.07, 179.33], [514.84, 179.31]]}, {\"category\": \"car\", \"corners_3d\": [[596.2, 202.06], [622.41, 202.0], [626.94, 204.37], [598.59, 204.44], [596.2, 176.97], [622.41, 176.96], [626.94, 177.3], [598.59, 177.31]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[591.37, 202.62], [598.35, 202.39], [602.94, 202.63], [595.93, 202.86], [591.37, 172.85], [598.35, 172.85], [602.94, 172.85], [595.93, 172.85]]}]\n```" - } - ] - }, - { - "id": 595, - "data_source": "Mono3DRefer", - "scene_name": "006468", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-36.31, 294.83], [56.77, 277.28], [334.48, 277.34], [288.03, 294.91], [-36.31, 184.88], [56.77, 183.15], [334.48, 183.15], [288.03, 184.88]]}, {\"category\": \"car\", \"corners_3d\": [[206.2, 255.6], [241.53, 246.53], [443.54, 244.35], [433.41, 252.85], [206.2, 164.76], [241.53, 165.65], [443.54, 165.86], [433.41, 165.03]]}, {\"category\": \"car\", \"corners_3d\": [[751.06, 236.15], [693.87, 236.31], [677.51, 226.56], [725.96, 226.45], [751.06, 182.04], [693.87, 182.07], [677.51, 180.65], [725.96, 180.63]]}, {\"category\": \"car\", \"corners_3d\": [[402.24, 213.12], [413.15, 211.15], [488.55, 211.24], [481.5, 213.22], [402.24, 175.02], [413.15, 174.91], [488.55, 174.91], [481.5, 175.02]]}, {\"category\": \"car\", \"corners_3d\": [[517.87, 210.06], [512.65, 212.02], [422.3, 211.97], [432.04, 210.01], [517.87, 174.54], [512.65, 174.63], [422.3, 174.62], [432.04, 174.54]]}]\n```" - } - ] - }, - { - "id": 596, - "data_source": "Mono3DRefer", - "scene_name": "006468", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[692.53, 238.4], [773.37, 238.41], [860.96, 273.58], [736.74, 273.57], [692.53, 142.33], [773.37, 142.32], [860.96, 125.94], [736.74, 125.95]]}, {\"category\": \"van\", \"corners_3d\": [[649.03, 205.26], [689.81, 205.24], [710.02, 212.77], [659.77, 212.8], [649.03, 152.54], [689.81, 152.55], [710.02, 147.83], [659.77, 147.81]]}, {\"category\": \"van\", \"corners_3d\": [[561.54, 196.55], [559.56, 197.4], [479.55, 197.38], [484.31, 196.53], [561.54, 158.47], [559.56, 157.95], [479.55, 157.97], [484.31, 158.48]]}]\n```" - } - ] - }, - { - "id": 597, - "data_source": "Mono3DRefer", - "scene_name": "006468", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[692.53, 238.4], [773.37, 238.41], [860.96, 273.58], [736.74, 273.57], [692.53, 142.33], [773.37, 142.32], [860.96, 125.94], [736.74, 125.95]]}, {\"category\": \"car\", \"corners_3d\": [[-36.31, 294.83], [56.77, 277.28], [334.48, 277.34], [288.03, 294.91], [-36.31, 184.88], [56.77, 183.15], [334.48, 183.15], [288.03, 184.88]]}, {\"category\": \"car\", \"corners_3d\": [[206.2, 255.6], [241.53, 246.53], [443.54, 244.35], [433.41, 252.85], [206.2, 164.76], [241.53, 165.65], [443.54, 165.86], [433.41, 165.03]]}, {\"category\": \"car\", \"corners_3d\": [[751.06, 236.15], [693.87, 236.31], [677.51, 226.56], [725.96, 226.45], [751.06, 182.04], [693.87, 182.07], [677.51, 180.65], [725.96, 180.63]]}, {\"category\": \"van\", \"corners_3d\": [[649.03, 205.26], [689.81, 205.24], [710.02, 212.77], [659.77, 212.8], [649.03, 152.54], [689.81, 152.55], [710.02, 147.83], [659.77, 147.81]]}, {\"category\": \"car\", \"corners_3d\": [[402.24, 213.12], [413.15, 211.15], [488.55, 211.24], [481.5, 213.22], [402.24, 175.02], [413.15, 174.91], [488.55, 174.91], [481.5, 175.02]]}, {\"category\": \"car\", \"corners_3d\": [[517.87, 210.06], [512.65, 212.02], [422.3, 211.97], [432.04, 210.01], [517.87, 174.54], [512.65, 174.63], [422.3, 174.62], [432.04, 174.54]]}, {\"category\": \"van\", \"corners_3d\": [[561.54, 196.55], [559.56, 197.4], [479.55, 197.38], [484.31, 196.53], [561.54, 158.47], [559.56, 157.95], [479.55, 157.97], [484.31, 158.48]]}]\n```" - } - ] - }, - { - "id": 598, - "data_source": "Mono3DRefer", - "scene_name": "006516", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1101.76, 237.0], [1173.52, 240.73], [1152.87, 250.97], [1072.24, 246.08], [1101.76, 176.49], [1173.52, 176.7], [1152.87, 177.28], [1072.24, 177.0]]}, {\"category\": \"car\", \"corners_3d\": [[1036.42, 208.65], [998.97, 207.52], [1022.19, 204.8], [1057.3, 205.75], [1036.42, 178.47], [998.97, 178.29], [1022.19, 177.86], [1057.3, 178.01]]}]\n```" - } - ] - }, - { - "id": 599, - "data_source": "Mono3DRefer", - "scene_name": "006537", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[960.48, 236.77], [1030.39, 239.8], [1022.19, 252.73], [939.94, 248.45], [960.48, 172.45], [1030.39, 172.43], [1022.19, 172.35], [939.94, 172.38]]}, {\"category\": \"car\", \"corners_3d\": [[873.99, 208.47], [838.73, 207.65], [855.95, 204.38], [888.19, 205.06], [873.99, 178.09], [838.73, 177.97], [855.95, 177.48], [888.19, 177.58]]}]\n```" - } - ] - }, - { - "id": 600, - "data_source": "Mono3DRefer", - "scene_name": "006563", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1003.9, 237.2], [1073.54, 240.49], [1060.43, 252.46], [979.91, 247.94], [1003.9, 174.72], [1073.54, 174.82], [1060.43, 175.17], [979.91, 175.03]]}, {\"category\": \"car\", \"corners_3d\": [[926.1, 208.35], [890.07, 207.47], [906.55, 204.32], [939.61, 205.05], [926.1, 178.07], [890.07, 177.94], [906.55, 177.48], [939.61, 177.58]]}]\n```" - } - ] - }, - { - "id": 601, - "data_source": "Mono3DRefer", - "scene_name": "006576", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[472.05, 237.63], [422.43, 238.03], [438.92, 229.18], [481.75, 228.88], [472.05, 197.49], [422.43, 197.65], [438.92, 194.28], [481.75, 194.17]]}, {\"category\": \"car\", \"corners_3d\": [[490.86, 221.29], [449.85, 221.54], [460.87, 215.83], [497.05, 215.64], [490.86, 181.41], [449.85, 181.45], [460.87, 180.44], [497.05, 180.41]]}, {\"category\": \"car\", \"corners_3d\": [[822.07, 259.42], [733.54, 260.64], [683.93, 239.32], [751.71, 238.62], [822.07, 164.92], [733.54, 164.8], [683.93, 166.76], [751.71, 166.82]]}, {\"category\": \"car\", \"corners_3d\": [[415.36, 271.11], [337.61, 272.05], [383.64, 249.91], [443.82, 249.34], [415.36, 197.66], [337.61, 197.9], [383.64, 192.31], [443.82, 192.16]]}]\n```" - } - ] - }, - { - "id": 602, - "data_source": "Mono3DRefer", - "scene_name": "006588", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[929.64, 334.6], [1128.22, 340.13], [1955.13, 666.56], [1328.75, 621.29], [929.64, 168.35], [1128.22, 168.2], [1955.13, 159.11], [1328.75, 160.37]]}, {\"category\": \"car\", \"corners_3d\": [[-117.75, 318.79], [8.3, 297.39], [222.56, 305.41], [126.51, 329.92], [-117.75, 201.54], [8.3, 197.33], [222.56, 198.91], [126.51, 203.73]]}, {\"category\": \"car\", \"corners_3d\": [[580.47, 229.24], [642.22, 229.49], [641.27, 242.82], [565.06, 242.45], [580.47, 172.84], [642.22, 172.84], [641.27, 172.84], [565.06, 172.84]]}, {\"category\": \"car\", \"corners_3d\": [[778.84, 233.63], [842.76, 233.74], [903.09, 250.58], [821.4, 250.4], [778.84, 172.07], [842.76, 172.06], [903.09, 171.85], [821.4, 171.85]]}, {\"category\": \"car\", \"corners_3d\": [[757.82, 214.15], [810.08, 214.39], [840.87, 223.39], [777.15, 223.04], [757.82, 171.42], [810.08, 171.41], [840.87, 171.1], [777.15, 171.11]]}, {\"category\": \"car\", \"corners_3d\": [[635.49, 206.21], [668.91, 206.42], [664.41, 209.94], [627.56, 209.67], [635.49, 174.98], [668.91, 175.0], [664.41, 175.22], [627.56, 175.21]]}, {\"category\": \"car\", \"corners_3d\": [[746.38, 201.31], [782.61, 201.53], [790.67, 205.66], [749.21, 205.37], [746.38, 164.37], [782.61, 164.3], [790.67, 163.07], [749.21, 163.16]]}]\n```" - } - ] - }, - { - "id": 603, - "data_source": "Mono3DRefer", - "scene_name": "006595", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[-209.44, 468.25], [-455.3, 469.33], [2.88, 340.82], [141.45, 340.48], [-209.44, 245.09], [-455.3, 245.35], [2.88, 213.93], [141.45, 213.84]]}, {\"category\": \"car\", \"corners_3d\": [[746.4, 268.97], [854.97, 268.98], [972.84, 315.25], [811.99, 315.23], [746.4, 175.31], [854.97, 175.31], [972.84, 176.5], [811.99, 176.5]]}, {\"category\": \"car\", \"corners_3d\": [[232.46, 306.64], [123.93, 306.45], [265.18, 268.12], [342.68, 268.21], [232.46, 208.27], [123.93, 208.22], [265.18, 198.07], [342.68, 198.1]]}, {\"category\": \"car\", \"corners_3d\": [[408.58, 249.84], [343.07, 249.85], [392.11, 235.65], [445.53, 235.65], [408.58, 180.78], [343.07, 180.78], [392.11, 179.32], [445.53, 179.32]]}, {\"category\": \"car\", \"corners_3d\": [[573.59, 199.46], [546.72, 199.44], [553.26, 197.17], [577.84, 197.19], [573.59, 177.89], [546.72, 177.89], [553.26, 177.46], [577.84, 177.46]]}, {\"category\": \"car\", \"corners_3d\": [[455.58, 225.93], [407.87, 225.9], [436.85, 218.52], [477.94, 218.55], [455.58, 178.97], [407.87, 178.97], [436.85, 178.12], [477.94, 178.12]]}, {\"category\": \"car\", \"corners_3d\": [[502.76, 211.69], [468.36, 211.69], [481.71, 208.01], [512.84, 208.01], [502.76, 182.76], [468.36, 182.76], [481.71, 181.82], [512.84, 181.82]]}, {\"category\": \"car\", \"corners_3d\": [[638.94, 192.53], [658.51, 192.54], [661.51, 193.97], [640.5, 193.96], [638.94, 174.26], [658.51, 174.26], [661.51, 174.37], [640.5, 174.37]]}, {\"category\": \"car\", \"corners_3d\": [[700.09, 188.8], [697.21, 188.29], [760.76, 188.29], [765.75, 188.8], [700.09, 168.05], [697.21, 168.2], [760.76, 168.2], [765.75, 168.05]]}]\n```" - } - ] - }, - { - "id": 604, - "data_source": "Mono3DRefer", - "scene_name": "006618", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[865.92, 401.14], [1072.54, 398.54], [2443.6, 1028.37], [1696.2, 1066.97], [865.92, 190.37], [1072.54, 190.17], [2443.6, 238.5], [1696.2, 241.46]]}, {\"category\": \"car\", \"corners_3d\": [[319.88, 223.39], [338.35, 220.4], [428.63, 220.59], [415.77, 223.6], [319.88, 176.89], [338.35, 176.65], [428.63, 176.66], [415.77, 176.9]]}, {\"category\": \"car\", \"corners_3d\": [[739.89, 202.22], [706.77, 202.3], [691.95, 199.21], [721.63, 199.14], [739.89, 167.4], [706.77, 167.39], [691.95, 167.96], [721.63, 167.97]]}, {\"category\": \"car\", \"corners_3d\": [[588.13, 205.11], [560.12, 205.16], [562.04, 202.07], [587.36, 202.02], [588.13, 177.91], [560.12, 177.92], [562.04, 177.43], [587.36, 177.43]]}]\n```" - } - ] - }, - { - "id": 605, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[685.66, 225.7], [683.04, 224.72], [696.49, 224.64], [699.35, 225.6], [685.66, 170.06], [683.04, 170.11], [696.49, 170.12], [699.35, 170.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[668.86, 219.81], [661.03, 219.12], [669.79, 218.73], [677.69, 219.41], [668.86, 171.56], [661.03, 171.57], [669.79, 171.59], [677.69, 171.57]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[413.85, 229.7], [410.57, 230.77], [399.76, 230.8], [403.23, 229.73], [413.85, 173.54], [410.57, 173.56], [399.76, 173.56], [403.23, 173.54]]}]\n```" - } - ] - }, - { - "id": 606, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[589.47, 199.19], [614.32, 199.18], [615.83, 201.31], [588.96, 201.33], [589.47, 174.63], [614.32, 174.63], [615.83, 174.78], [588.96, 174.78]]}, {\"category\": \"car\", \"corners_3d\": [[449.35, 251.27], [518.2, 251.27], [481.66, 282.44], [385.46, 282.43], [449.35, 177.85], [518.2, 177.85], [481.66, 179.84], [385.46, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[200.49, 249.39], [138.25, 249.65], [205.31, 238.0], [257.93, 237.81], [200.49, 192.6], [138.25, 192.66], [205.31, 189.66], [257.93, 189.61]]}]\n```" - } - ] - }, - { - "id": 607, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[676.07, 225.02], [686.98, 224.16], [735.98, 227.34], [725.25, 228.31], [676.07, 162.26], [686.98, 162.43], [735.98, 161.79], [725.25, 161.59]]}]\n```" - } - ] - }, - { - "id": 608, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[589.47, 199.19], [614.32, 199.18], [615.83, 201.31], [588.96, 201.33], [589.47, 174.63], [614.32, 174.63], [615.83, 174.78], [588.96, 174.78]]}, {\"category\": \"car\", \"corners_3d\": [[449.35, 251.27], [518.2, 251.27], [481.66, 282.44], [385.46, 282.43], [449.35, 177.85], [518.2, 177.85], [481.66, 179.84], [385.46, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[200.49, 249.39], [138.25, 249.65], [205.31, 238.0], [257.93, 237.81], [200.49, 192.6], [138.25, 192.66], [205.31, 189.66], [257.93, 189.61]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[685.66, 225.7], [683.04, 224.72], [696.49, 224.64], [699.35, 225.6], [685.66, 170.06], [683.04, 170.11], [696.49, 170.12], [699.35, 170.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[668.86, 219.81], [661.03, 219.12], [669.79, 218.73], [677.69, 219.41], [668.86, 171.56], [661.03, 171.57], [669.79, 171.59], [677.69, 171.57]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[413.85, 229.7], [410.57, 230.77], [399.76, 230.8], [403.23, 229.73], [413.85, 173.54], [410.57, 173.56], [399.76, 173.56], [403.23, 173.54]]}]\n```" - } - ] - }, - { - "id": 609, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[676.07, 225.02], [686.98, 224.16], [735.98, 227.34], [725.25, 228.31], [676.07, 162.26], [686.98, 162.43], [735.98, 161.79], [725.25, 161.59]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[685.66, 225.7], [683.04, 224.72], [696.49, 224.64], [699.35, 225.6], [685.66, 170.06], [683.04, 170.11], [696.49, 170.12], [699.35, 170.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[668.86, 219.81], [661.03, 219.12], [669.79, 218.73], [677.69, 219.41], [668.86, 171.56], [661.03, 171.57], [669.79, 171.59], [677.69, 171.57]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[413.85, 229.7], [410.57, 230.77], [399.76, 230.8], [403.23, 229.73], [413.85, 173.54], [410.57, 173.56], [399.76, 173.56], [403.23, 173.54]]}]\n```" - } - ] - }, - { - "id": 610, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[589.47, 199.19], [614.32, 199.18], [615.83, 201.31], [588.96, 201.33], [589.47, 174.63], [614.32, 174.63], [615.83, 174.78], [588.96, 174.78]]}, {\"category\": \"car\", \"corners_3d\": [[449.35, 251.27], [518.2, 251.27], [481.66, 282.44], [385.46, 282.43], [449.35, 177.85], [518.2, 177.85], [481.66, 179.84], [385.46, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[200.49, 249.39], [138.25, 249.65], [205.31, 238.0], [257.93, 237.81], [200.49, 192.6], [138.25, 192.66], [205.31, 189.66], [257.93, 189.61]]}, {\"category\": \"cyclist\", \"corners_3d\": [[676.07, 225.02], [686.98, 224.16], [735.98, 227.34], [725.25, 228.31], [676.07, 162.26], [686.98, 162.43], [735.98, 161.79], [725.25, 161.59]]}]\n```" - } - ] - }, - { - "id": 611, - "data_source": "Mono3DRefer", - "scene_name": "006623", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[589.47, 199.19], [614.32, 199.18], [615.83, 201.31], [588.96, 201.33], [589.47, 174.63], [614.32, 174.63], [615.83, 174.78], [588.96, 174.78]]}, {\"category\": \"car\", \"corners_3d\": [[449.35, 251.27], [518.2, 251.27], [481.66, 282.44], [385.46, 282.43], [449.35, 177.85], [518.2, 177.85], [481.66, 179.84], [385.46, 179.84]]}, {\"category\": \"car\", \"corners_3d\": [[200.49, 249.39], [138.25, 249.65], [205.31, 238.0], [257.93, 237.81], [200.49, 192.6], [138.25, 192.66], [205.31, 189.66], [257.93, 189.61]]}, {\"category\": \"cyclist\", \"corners_3d\": [[676.07, 225.02], [686.98, 224.16], [735.98, 227.34], [725.25, 228.31], [676.07, 162.26], [686.98, 162.43], [735.98, 161.79], [725.25, 161.59]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[685.66, 225.7], [683.04, 224.72], [696.49, 224.64], [699.35, 225.6], [685.66, 170.06], [683.04, 170.11], [696.49, 170.12], [699.35, 170.06]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[668.86, 219.81], [661.03, 219.12], [669.79, 218.73], [677.69, 219.41], [668.86, 171.56], [661.03, 171.57], [669.79, 171.59], [677.69, 171.57]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[413.85, 229.7], [410.57, 230.77], [399.76, 230.8], [403.23, 229.73], [413.85, 173.54], [410.57, 173.56], [399.76, 173.56], [403.23, 173.54]]}]\n```" - } - ] - }, - { - "id": 612, - "data_source": "Mono3DRefer", - "scene_name": "006677", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[517.13, 244.81], [463.19, 244.92], [484.43, 233.27], [529.63, 233.19], [517.13, 195.06], [463.19, 195.1], [484.43, 191.5], [529.63, 191.48]]}]\n```" - } - ] - }, - { - "id": 613, - "data_source": "Mono3DRefer", - "scene_name": "006678", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[636.29, 210.66], [670.62, 210.6], [681.8, 215.74], [642.81, 215.81], [636.29, 182.77], [670.62, 182.76], [681.8, 184.11], [642.81, 184.13]]}, {\"category\": \"car\", \"corners_3d\": [[490.73, 222.73], [483.24, 225.87], [359.4, 225.87], [374.23, 222.73], [490.73, 179.42], [483.24, 179.84], [359.4, 179.84], [374.23, 179.42]]}, {\"category\": \"car\", \"corners_3d\": [[241.52, 252.67], [273.04, 245.67], [428.75, 245.49], [412.23, 252.45], [241.52, 186.87], [273.04, 185.65], [428.75, 185.61], [412.23, 186.84]]}, {\"category\": \"car\", \"corners_3d\": [[907.11, 240.61], [938.56, 247.14], [773.57, 247.83], [756.48, 241.18], [907.11, 181.85], [938.56, 182.72], [773.57, 182.81], [756.48, 181.93]]}, {\"category\": \"car\", \"corners_3d\": [[166.48, 262.06], [210.79, 253.16], [393.63, 253.19], [369.57, 262.1], [166.48, 187.79], [210.79, 186.3], [393.63, 186.31], [369.57, 187.8]]}, {\"category\": \"car\", \"corners_3d\": [[63.5, 284.8], [127.45, 270.31], [361.99, 268.15], [333.64, 281.95], [63.5, 171.55], [127.45, 171.72], [361.99, 171.75], [333.64, 171.58]]}, {\"category\": \"car\", \"corners_3d\": [[330.67, 293.47], [279.07, 315.23], [-67.9, 314.54], [36.28, 292.98], [330.67, 188.78], [279.07, 191.65], [-67.9, 191.56], [36.28, 188.71]]}, {\"category\": \"car\", \"corners_3d\": [[267.18, 331.25], [181.89, 367.64], [-227.55, 363.11], [-68.86, 328.24], [267.18, 195.45], [181.89, 200.65], [-227.55, 200.0], [-68.86, 195.03]]}, {\"category\": \"car\", \"corners_3d\": [[1275.95, 310.64], [1433.62, 340.94], [1097.86, 346.9], [997.91, 314.62], [1275.95, 182.73], [1433.62, 184.9], [1097.86, 185.33], [997.91, 183.01]]}, {\"category\": \"car\", \"corners_3d\": [[1507.52, 350.48], [1819.8, 407.64], [1363.55, 425.55], [1150.95, 360.54], [1507.52, 195.02], [1819.8, 202.16], [1363.55, 204.39], [1150.95, 196.28]]}]\n```" - } - ] - }, - { - "id": 614, - "data_source": "Mono3DRefer", - "scene_name": "006732", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[510.24, 311.9], [377.61, 311.41], [451.31, 269.87], [544.26, 270.11], [510.24, 196.8], [377.61, 196.72], [451.31, 189.56], [544.26, 189.61]]}]\n```" - } - ] - }, - { - "id": 615, - "data_source": "Mono3DRefer", - "scene_name": "006772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[798.39, 317.0], [924.79, 317.02], [1159.98, 424.78], [939.07, 424.71], [798.39, 194.69], [924.79, 194.7], [1159.98, 211.02], [939.07, 211.01]]}, {\"category\": \"car\", \"corners_3d\": [[721.98, 262.64], [807.23, 262.43], [885.53, 295.59], [768.87, 295.98], [721.98, 182.53], [807.23, 182.51], [885.53, 186.08], [768.87, 186.12]]}, {\"category\": \"car\", \"corners_3d\": [[536.25, 228.09], [487.9, 228.35], [498.18, 220.25], [539.46, 220.06], [536.25, 179.55], [487.9, 179.58], [498.18, 178.6], [539.46, 178.58]]}, {\"category\": \"car\", \"corners_3d\": [[659.93, 212.11], [695.36, 212.09], [707.22, 217.12], [667.25, 217.15], [659.93, 180.21], [695.36, 180.2], [707.22, 181.15], [667.25, 181.15]]}, {\"category\": \"car\", \"corners_3d\": [[556.47, 208.14], [523.93, 208.19], [530.55, 204.5], [559.69, 204.46], [556.47, 181.82], [523.93, 181.84], [530.55, 180.9], [559.69, 180.89]]}]\n```" - } - ] - }, - { - "id": 616, - "data_source": "Mono3DRefer", - "scene_name": "006772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[673.57, 230.83], [737.49, 230.73], [785.21, 250.21], [699.85, 250.39], [673.57, 153.73], [737.49, 153.76], [785.21, 147.34], [699.85, 147.28]]}]\n```" - } - ] - }, - { - "id": 617, - "data_source": "Mono3DRefer", - "scene_name": "006772", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[798.39, 317.0], [924.79, 317.02], [1159.98, 424.78], [939.07, 424.71], [798.39, 194.69], [924.79, 194.7], [1159.98, 211.02], [939.07, 211.01]]}, {\"category\": \"car\", \"corners_3d\": [[721.98, 262.64], [807.23, 262.43], [885.53, 295.59], [768.87, 295.98], [721.98, 182.53], [807.23, 182.51], [885.53, 186.08], [768.87, 186.12]]}, {\"category\": \"van\", \"corners_3d\": [[673.57, 230.83], [737.49, 230.73], [785.21, 250.21], [699.85, 250.39], [673.57, 153.73], [737.49, 153.76], [785.21, 147.34], [699.85, 147.28]]}, {\"category\": \"car\", \"corners_3d\": [[536.25, 228.09], [487.9, 228.35], [498.18, 220.25], [539.46, 220.06], [536.25, 179.55], [487.9, 179.58], [498.18, 178.6], [539.46, 178.58]]}, {\"category\": \"car\", \"corners_3d\": [[659.93, 212.11], [695.36, 212.09], [707.22, 217.12], [667.25, 217.15], [659.93, 180.21], [695.36, 180.2], [707.22, 181.15], [667.25, 181.15]]}, {\"category\": \"car\", \"corners_3d\": [[556.47, 208.14], [523.93, 208.19], [530.55, 204.5], [559.69, 204.46], [556.47, 181.82], [523.93, 181.84], [530.55, 180.9], [559.69, 180.89]]}]\n```" - } - ] - }, - { - "id": 618, - "data_source": "Mono3DRefer", - "scene_name": "006797", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[552.01, 224.22], [507.41, 223.96], [526.73, 217.77], [565.99, 217.96], [552.01, 184.29], [507.41, 184.23], [526.73, 182.85], [565.99, 182.9]]}]\n```" - } - ] - }, - { - "id": 619, - "data_source": "Mono3DRefer", - "scene_name": "006806", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[431.77, 283.18], [339.21, 283.75], [397.97, 256.7], [467.81, 256.37], [431.77, 197.49], [339.21, 197.62], [397.97, 191.58], [467.81, 191.5]]}, {\"category\": \"car\", \"corners_3d\": [[555.74, 208.03], [519.86, 208.06], [527.86, 204.29], [559.89, 204.26], [555.74, 178.16], [519.86, 178.17], [527.86, 177.6], [559.89, 177.6]]}]\n```" - } - ] - }, - { - "id": 620, - "data_source": "Mono3DRefer", - "scene_name": "006806", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[489.9, 244.6], [421.64, 244.67], [455.37, 231.22], [510.83, 231.17], [489.9, 170.38], [421.64, 170.38], [455.37, 170.84], [510.83, 170.84]]}]\n```" - } - ] - }, - { - "id": 621, - "data_source": "Mono3DRefer", - "scene_name": "006806", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[431.77, 283.18], [339.21, 283.75], [397.97, 256.7], [467.81, 256.37], [431.77, 197.49], [339.21, 197.62], [397.97, 191.58], [467.81, 191.5]]}, {\"category\": \"van\", \"corners_3d\": [[489.9, 244.6], [421.64, 244.67], [455.37, 231.22], [510.83, 231.17], [489.9, 170.38], [421.64, 170.38], [455.37, 170.84], [510.83, 170.84]]}, {\"category\": \"car\", \"corners_3d\": [[555.74, 208.03], [519.86, 208.06], [527.86, 204.29], [559.89, 204.26], [555.74, 178.16], [519.86, 178.17], [527.86, 177.6], [559.89, 177.6]]}]\n```" - } - ] - }, - { - "id": 622, - "data_source": "Mono3DRefer", - "scene_name": "006816", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[654.45, 207.01], [679.68, 207.02], [687.1, 210.66], [659.18, 210.66], [654.45, 184.5], [679.68, 184.5], [687.1, 185.75], [659.18, 185.75]]}]\n```" - } - ] - }, - { - "id": 623, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[1077.35, 372.17], [1027.68, 352.28], [1079.29, 351.74], [1134.48, 371.49], [1077.35, 123.66], [1027.68, 128.57], [1079.29, 128.71], [1134.48, 123.83]]}]\n```" - } - ] - }, - { - "id": 624, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[450.62, 258.38], [531.34, 258.48], [499.81, 289.87], [389.61, 289.68], [450.62, 195.52], [531.34, 195.55], [499.81, 203.87], [389.61, 203.82]]}, {\"category\": \"car\", \"corners_3d\": [[665.1, 214.9], [703.37, 214.86], [722.03, 222.14], [677.15, 222.2], [665.1, 178.26], [703.37, 178.25], [722.03, 179.19], [677.15, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[511.34, 229.65], [556.77, 229.73], [542.88, 241.38], [488.18, 241.27], [511.34, 187.42], [556.77, 187.44], [542.88, 190.42], [488.18, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[704.24, 245.26], [772.79, 245.26], [822.04, 267.19], [732.73, 267.18], [704.24, 176.72], [772.79, 176.72], [822.04, 177.89], [732.73, 177.89]]}, {\"category\": \"car\", \"corners_3d\": [[647.68, 201.34], [673.6, 201.43], [673.46, 203.79], [645.41, 203.69], [647.68, 179.62], [673.6, 179.64], [673.46, 180.2], [645.41, 180.17]]}]\n```" - } - ] - }, - { - "id": 625, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[731.35, 262.78], [840.24, 262.65], [1097.9, 360.14], [871.19, 360.69], [731.35, 112.09], [840.24, 112.18], [1097.9, 46.3], [871.19, 45.93]]}]\n```" - } - ] - }, - { - "id": 626, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[450.62, 258.38], [531.34, 258.48], [499.81, 289.87], [389.61, 289.68], [450.62, 195.52], [531.34, 195.55], [499.81, 203.87], [389.61, 203.82]]}, {\"category\": \"car\", \"corners_3d\": [[665.1, 214.9], [703.37, 214.86], [722.03, 222.14], [677.15, 222.2], [665.1, 178.26], [703.37, 178.25], [722.03, 179.19], [677.15, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[511.34, 229.65], [556.77, 229.73], [542.88, 241.38], [488.18, 241.27], [511.34, 187.42], [556.77, 187.44], [542.88, 190.42], [488.18, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[704.24, 245.26], [772.79, 245.26], [822.04, 267.19], [732.73, 267.18], [704.24, 176.72], [772.79, 176.72], [822.04, 177.89], [732.73, 177.89]]}, {\"category\": \"car\", \"corners_3d\": [[647.68, 201.34], [673.6, 201.43], [673.46, 203.79], [645.41, 203.69], [647.68, 179.62], [673.6, 179.64], [673.46, 180.2], [645.41, 180.17]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1077.35, 372.17], [1027.68, 352.28], [1079.29, 351.74], [1134.48, 371.49], [1077.35, 123.66], [1027.68, 128.57], [1079.29, 128.71], [1134.48, 123.83]]}]\n```" - } - ] - }, - { - "id": 627, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[731.35, 262.78], [840.24, 262.65], [1097.9, 360.14], [871.19, 360.69], [731.35, 112.09], [840.24, 112.18], [1097.9, 46.3], [871.19, 45.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1077.35, 372.17], [1027.68, 352.28], [1079.29, 351.74], [1134.48, 371.49], [1077.35, 123.66], [1027.68, 128.57], [1079.29, 128.71], [1134.48, 123.83]]}]\n```" - } - ] - }, - { - "id": 628, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[450.62, 258.38], [531.34, 258.48], [499.81, 289.87], [389.61, 289.68], [450.62, 195.52], [531.34, 195.55], [499.81, 203.87], [389.61, 203.82]]}, {\"category\": \"van\", \"corners_3d\": [[731.35, 262.78], [840.24, 262.65], [1097.9, 360.14], [871.19, 360.69], [731.35, 112.09], [840.24, 112.18], [1097.9, 46.3], [871.19, 45.93]]}, {\"category\": \"car\", \"corners_3d\": [[665.1, 214.9], [703.37, 214.86], [722.03, 222.14], [677.15, 222.2], [665.1, 178.26], [703.37, 178.25], [722.03, 179.19], [677.15, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[511.34, 229.65], [556.77, 229.73], [542.88, 241.38], [488.18, 241.27], [511.34, 187.42], [556.77, 187.44], [542.88, 190.42], [488.18, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[704.24, 245.26], [772.79, 245.26], [822.04, 267.19], [732.73, 267.18], [704.24, 176.72], [772.79, 176.72], [822.04, 177.89], [732.73, 177.89]]}, {\"category\": \"car\", \"corners_3d\": [[647.68, 201.34], [673.6, 201.43], [673.46, 203.79], [645.41, 203.69], [647.68, 179.62], [673.6, 179.64], [673.46, 180.2], [645.41, 180.17]]}]\n```" - } - ] - }, - { - "id": 629, - "data_source": "Mono3DRefer", - "scene_name": "006822", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian, car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[450.62, 258.38], [531.34, 258.48], [499.81, 289.87], [389.61, 289.68], [450.62, 195.52], [531.34, 195.55], [499.81, 203.87], [389.61, 203.82]]}, {\"category\": \"van\", \"corners_3d\": [[731.35, 262.78], [840.24, 262.65], [1097.9, 360.14], [871.19, 360.69], [731.35, 112.09], [840.24, 112.18], [1097.9, 46.3], [871.19, 45.93]]}, {\"category\": \"car\", \"corners_3d\": [[665.1, 214.9], [703.37, 214.86], [722.03, 222.14], [677.15, 222.2], [665.1, 178.26], [703.37, 178.25], [722.03, 179.19], [677.15, 179.19]]}, {\"category\": \"car\", \"corners_3d\": [[511.34, 229.65], [556.77, 229.73], [542.88, 241.38], [488.18, 241.27], [511.34, 187.42], [556.77, 187.44], [542.88, 190.42], [488.18, 190.4]]}, {\"category\": \"car\", \"corners_3d\": [[704.24, 245.26], [772.79, 245.26], [822.04, 267.19], [732.73, 267.18], [704.24, 176.72], [772.79, 176.72], [822.04, 177.89], [732.73, 177.89]]}, {\"category\": \"car\", \"corners_3d\": [[647.68, 201.34], [673.6, 201.43], [673.46, 203.79], [645.41, 203.69], [647.68, 179.62], [673.6, 179.64], [673.46, 180.2], [645.41, 180.17]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1077.35, 372.17], [1027.68, 352.28], [1079.29, 351.74], [1134.48, 371.49], [1077.35, 123.66], [1027.68, 128.57], [1079.29, 128.71], [1134.48, 123.83]]}]\n```" - } - ] - }, - { - "id": 630, - "data_source": "Mono3DRefer", - "scene_name": "006828", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[567.17, 219.63], [545.25, 219.65], [548.79, 216.79], [569.07, 216.78], [567.17, 163.54], [545.25, 163.53], [548.79, 164.88], [569.07, 164.88]]}, {\"category\": \"cyclist\", \"corners_3d\": [[381.69, 362.65], [258.95, 369.72], [295.95, 327.54], [390.89, 323.24], [381.69, 159.43], [258.95, 158.56], [295.95, 163.71], [390.89, 164.23]]}]\n```" - } - ] - }, - { - "id": 631, - "data_source": "Mono3DRefer", - "scene_name": "006828", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[439.83, 242.83], [446.04, 240.2], [484.27, 240.13], [479.77, 242.75], [439.83, 159.86], [446.04, 160.79], [484.27, 160.81], [479.77, 159.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[556.11, 233.3], [523.09, 233.21], [528.07, 230.77], [559.54, 230.85], [556.11, 157.81], [523.09, 157.85], [528.07, 158.97], [559.54, 158.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[759.01, 227.65], [724.36, 227.65], [716.9, 224.89], [749.48, 224.89], [759.01, 144.13], [724.36, 144.12], [716.9, 146.36], [749.48, 146.36]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1155.0, 326.05], [1260.37, 337.95], [1255.75, 351.89], [1142.14, 337.87], [1155.0, 138.98], [1260.37, 135.48], [1255.75, 131.38], [1142.14, 135.5]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[808.89, 254.86], [857.92, 254.56], [889.99, 262.37], [835.84, 262.74], [808.89, 147.76], [857.92, 147.9], [889.99, 144.31], [835.84, 144.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[571.79, 240.86], [613.7, 241.72], [603.61, 245.45], [559.29, 244.48], [571.79, 159.96], [613.7, 159.65], [603.61, 158.3], [559.29, 158.65]]}]\n```" - } - ] - }, - { - "id": 632, - "data_source": "Mono3DRefer", - "scene_name": "006828", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist, pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.34, 718.34), (cx, cy) = (600.39, 181.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[439.83, 242.83], [446.04, 240.2], [484.27, 240.13], [479.77, 242.75], [439.83, 159.86], [446.04, 160.79], [484.27, 160.81], [479.77, 159.89]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[556.11, 233.3], [523.09, 233.21], [528.07, 230.77], [559.54, 230.85], [556.11, 157.81], [523.09, 157.85], [528.07, 158.97], [559.54, 158.93]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[759.01, 227.65], [724.36, 227.65], [716.9, 224.89], [749.48, 224.89], [759.01, 144.13], [724.36, 144.12], [716.9, 146.36], [749.48, 146.36]]}, {\"category\": \"cyclist\", \"corners_3d\": [[567.17, 219.63], [545.25, 219.65], [548.79, 216.79], [569.07, 216.78], [567.17, 163.54], [545.25, 163.53], [548.79, 164.88], [569.07, 164.88]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[1155.0, 326.05], [1260.37, 337.95], [1255.75, 351.89], [1142.14, 337.87], [1155.0, 138.98], [1260.37, 135.48], [1255.75, 131.38], [1142.14, 135.5]]}, {\"category\": \"cyclist\", \"corners_3d\": [[381.69, 362.65], [258.95, 369.72], [295.95, 327.54], [390.89, 323.24], [381.69, 159.43], [258.95, 158.56], [295.95, 163.71], [390.89, 164.23]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[808.89, 254.86], [857.92, 254.56], [889.99, 262.37], [835.84, 262.74], [808.89, 147.76], [857.92, 147.9], [889.99, 144.31], [835.84, 144.14]]}, {\"category\": \"pedestrian\", \"corners_3d\": [[571.79, 240.86], [613.7, 241.72], [603.61, 245.45], [559.29, 244.48], [571.79, 159.96], [613.7, 159.65], [603.61, 158.3], [559.29, 158.65]]}]\n```" - } - ] - }, - { - "id": 633, - "data_source": "Mono3DRefer", - "scene_name": "006850", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[943.44, 286.98], [1061.07, 306.15], [854.8, 334.52], [751.27, 307.16], [943.44, 168.38], [1061.07, 167.63], [854.8, 166.52], [751.27, 167.59]]}, {\"category\": \"car\", \"corners_3d\": [[1129.13, 302.26], [1008.91, 284.73], [1156.07, 268.52], [1275.87, 281.05], [1129.13, 169.44], [1008.91, 169.9], [1156.07, 170.33], [1275.87, 170.0]]}]\n```" - } - ] - }, - { - "id": 634, - "data_source": "Mono3DRefer", - "scene_name": "006866", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[891.02, 353.43], [1081.48, 352.02], [1864.42, 636.55], [1381.4, 646.11], [891.02, 188.72], [1081.48, 188.59], [1864.42, 213.59], [1381.4, 214.43]]}, {\"category\": \"car\", \"corners_3d\": [[-42.55, 627.76], [-470.14, 633.26], [150.19, 365.01], [326.79, 364.04], [-42.55, 192.87], [-470.14, 193.12], [150.19, 181.31], [326.79, 181.27]]}, {\"category\": \"car\", \"corners_3d\": [[725.19, 265.34], [830.16, 264.26], [973.76, 311.49], [816.15, 314.0], [725.19, 163.7], [830.16, 163.8], [973.76, 159.13], [816.15, 158.88]]}, {\"category\": \"car\", \"corners_3d\": [[366.34, 341.46], [225.91, 343.1], [328.0, 293.01], [426.69, 292.19], [366.34, 189.14], [225.91, 189.3], [328.0, 184.46], [426.69, 184.38]]}, {\"category\": \"car\", \"corners_3d\": [[444.44, 270.39], [362.25, 270.73], [410.84, 249.6], [475.2, 249.39], [444.44, 202.25], [362.25, 202.35], [410.84, 195.98], [475.2, 195.92]]}, {\"category\": \"car\", \"corners_3d\": [[513.34, 221.96], [478.67, 221.99], [491.73, 216.78], [522.72, 216.75], [513.34, 183.42], [478.67, 183.42], [491.73, 182.3], [522.72, 182.3]]}]\n```" - } - ] - }, - { - "id": 635, - "data_source": "Mono3DRefer", - "scene_name": "006898", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[143.91, 220.18], [106.84, 221.3], [109.13, 217.49], [143.3, 216.54], [143.91, 192.94], [106.84, 193.42], [109.13, 191.8], [143.3, 191.4]]}]\n```" - } - ] - }, - { - "id": 636, - "data_source": "Mono3DRefer", - "scene_name": "006900", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[508.29, 244.45], [450.62, 244.56], [475.22, 232.31], [523.02, 232.23], [508.29, 191.27], [450.62, 191.3], [475.22, 188.15], [523.02, 188.13]]}]\n```" - } - ] - }, - { - "id": 637, - "data_source": "Mono3DRefer", - "scene_name": "006978", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (718.86, 718.86), (cx, cy) = (607.19, 185.22). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[646.14, 231.56], [702.91, 231.93], [707.67, 241.59], [639.23, 241.06], [646.14, 180.38], [702.91, 180.34], [707.67, 179.34], [639.23, 179.39]]}, {\"category\": \"car\", \"corners_3d\": [[402.45, 229.68], [451.91, 229.53], [428.53, 238.17], [369.29, 238.38], [402.45, 183.08], [451.91, 183.08], [428.53, 182.67], [369.29, 182.66]]}, {\"category\": \"car\", \"corners_3d\": [[442.75, 217.03], [479.49, 216.96], [465.36, 221.47], [423.35, 221.56], [442.75, 184.72], [479.49, 184.72], [465.36, 184.65], [423.35, 184.65]]}, {\"category\": \"car\", \"corners_3d\": [[726.03, 205.59], [748.9, 205.81], [740.28, 207.27], [715.9, 207.02], [726.03, 185.34], [748.9, 185.35], [740.28, 185.36], [715.9, 185.35]]}, {\"category\": \"car\", \"corners_3d\": [[810.11, 198.63], [826.46, 198.84], [809.65, 199.43], [792.87, 199.2], [810.11, 184.18], [826.46, 184.16], [809.65, 184.12], [792.87, 184.13]]}]\n```" - } - ] - }, - { - "id": 638, - "data_source": "Mono3DRefer", - "scene_name": "006981", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[607.49, 244.52], [575.99, 244.59], [577.88, 237.31], [606.17, 237.26], [607.49, 215.39], [575.99, 215.43], [577.88, 211.11], [606.17, 211.08]]}]\n```" - } - ] - }, - { - "id": 639, - "data_source": "Mono3DRefer", - "scene_name": "006987", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1463.83, 394.84], [1843.14, 491.77], [1190.85, 497.75], [1007.44, 397.72], [1463.83, 195.68], [1843.14, 205.65], [1190.85, 206.26], [1007.44, 195.98]]}, {\"category\": \"car\", \"corners_3d\": [[-288.53, 316.82], [-152.46, 292.41], [174.51, 286.12], [109.54, 307.8], [-288.53, 160.9], [-152.46, 162.92], [174.51, 163.45], [109.54, 161.65]]}, {\"category\": \"car\", \"corners_3d\": [[264.03, 262.87], [225.12, 275.36], [-24.65, 278.41], [46.51, 265.22], [264.03, 151.2], [225.12, 148.2], [-24.65, 147.46], [46.51, 150.64]]}, {\"category\": \"car\", \"corners_3d\": [[130.06, 237.59], [158.97, 231.82], [317.78, 229.36], [305.0, 234.64], [130.06, 159.71], [158.97, 160.88], [317.78, 161.38], [305.0, 160.31]]}, {\"category\": \"car\", \"corners_3d\": [[180.93, 231.38], [201.3, 226.63], [337.77, 224.43], [329.83, 228.79], [180.93, 175.47], [201.3, 175.26], [337.77, 175.16], [329.83, 175.35]]}, {\"category\": \"car\", \"corners_3d\": [[467.49, 201.67], [494.68, 201.24], [511.93, 203.83], [482.5, 204.33], [467.49, 174.01], [494.68, 174.0], [511.93, 174.1], [482.5, 174.12]]}]\n```" - } - ] - }, - { - "id": 640, - "data_source": "Mono3DRefer", - "scene_name": "007048", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[1177.53, 235.98], [1251.56, 240.0], [1221.35, 248.73], [1140.36, 243.64], [1177.53, 176.43], [1251.56, 176.66], [1221.35, 177.15], [1140.36, 176.86]]}, {\"category\": \"car\", \"corners_3d\": [[1132.17, 211.58], [1091.4, 210.22], [1116.32, 207.46], [1154.84, 208.63], [1132.17, 180.55], [1091.4, 180.28], [1116.32, 179.73], [1154.84, 179.96]]}]\n```" - } - ] - }, - { - "id": 641, - "data_source": "Mono3DRefer", - "scene_name": "007187", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[341.25, 216.8], [308.06, 217.42], [312.78, 213.5], [343.03, 212.98], [341.25, 189.33], [308.06, 189.56], [312.78, 188.09], [343.03, 187.9]]}]\n```" - } - ] - }, - { - "id": 642, - "data_source": "Mono3DRefer", - "scene_name": "007212", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[656.4, 220.11], [688.84, 220.09], [701.01, 226.78], [663.99, 226.8], [656.4, 191.13], [688.84, 191.13], [701.01, 193.72], [663.99, 193.73]]}]\n```" - } - ] - }, - { - "id": 643, - "data_source": "Mono3DRefer", - "scene_name": "007214", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[759.25, 225.72], [784.53, 227.73], [718.81, 232.04], [694.22, 229.71], [759.25, 197.2], [784.53, 198.13], [718.81, 200.11], [694.22, 199.04]]}, {\"category\": \"car\", \"corners_3d\": [[1001.89, 241.16], [999.13, 235.51], [1207.79, 242.61], [1232.06, 249.69], [1001.89, 174.57], [999.13, 174.43], [1207.79, 174.6], [1232.06, 174.78]]}]\n```" - } - ] - }, - { - "id": 644, - "data_source": "Mono3DRefer", - "scene_name": "007214", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[604.54, 221.54], [642.68, 221.34], [655.94, 228.45], [612.25, 228.72], [604.54, 178.26], [642.68, 178.24], [655.94, 179.03], [612.25, 179.06]]}, {\"category\": \"van\", \"corners_3d\": [[522.53, 201.49], [543.79, 201.35], [547.89, 203.38], [525.13, 203.55], [522.53, 175.28], [543.79, 175.27], [547.89, 175.44], [525.13, 175.46]]}]\n```" - } - ] - }, - { - "id": 645, - "data_source": "Mono3DRefer", - "scene_name": "007214", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[604.54, 221.54], [642.68, 221.34], [655.94, 228.45], [612.25, 228.72], [604.54, 178.26], [642.68, 178.24], [655.94, 179.03], [612.25, 179.06]]}, {\"category\": \"car\", \"corners_3d\": [[759.25, 225.72], [784.53, 227.73], [718.81, 232.04], [694.22, 229.71], [759.25, 197.2], [784.53, 198.13], [718.81, 200.11], [694.22, 199.04]]}, {\"category\": \"van\", \"corners_3d\": [[522.53, 201.49], [543.79, 201.35], [547.89, 203.38], [525.13, 203.55], [522.53, 175.28], [543.79, 175.27], [547.89, 175.44], [525.13, 175.46]]}, {\"category\": \"car\", \"corners_3d\": [[1001.89, 241.16], [999.13, 235.51], [1207.79, 242.61], [1232.06, 249.69], [1001.89, 174.57], [999.13, 174.43], [1207.79, 174.6], [1232.06, 174.78]]}]\n```" - } - ] - }, - { - "id": 646, - "data_source": "Mono3DRefer", - "scene_name": "007232", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[195.13, 360.12], [9.3, 358.69], [254.67, 285.3], [367.91, 285.83], [195.13, 170.66], [9.3, 170.68], [254.67, 171.54], [367.91, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[437.22, 267.32], [355.93, 266.35], [430.66, 244.38], [493.3, 244.95], [437.22, 191.73], [355.93, 191.54], [430.66, 187.15], [493.3, 187.26]]}, {\"category\": \"car\", \"corners_3d\": [[584.53, 217.75], [546.04, 217.51], [560.26, 212.82], [594.76, 213.01], [584.53, 184.37], [546.04, 184.3], [560.26, 183.1], [594.76, 183.15]]}, {\"category\": \"car\", \"corners_3d\": [[659.53, 199.55], [638.48, 199.28], [652.38, 197.65], [672.25, 197.89], [659.53, 180.23], [638.48, 180.15], [652.38, 179.7], [672.25, 179.77]]}]\n```" - } - ] - }, - { - "id": 647, - "data_source": "Mono3DRefer", - "scene_name": "007232", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"cyclist\", \"corners_3d\": [[683.96, 233.62], [707.7, 233.78], [711.42, 239.48], [685.45, 239.29], [683.96, 172.84], [707.7, 172.84], [711.42, 172.84], [685.45, 172.84]]}]\n```" - } - ] - }, - { - "id": 648, - "data_source": "Mono3DRefer", - "scene_name": "007232", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car, cyclist). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[195.13, 360.12], [9.3, 358.69], [254.67, 285.3], [367.91, 285.83], [195.13, 170.66], [9.3, 170.68], [254.67, 171.54], [367.91, 171.53]]}, {\"category\": \"car\", \"corners_3d\": [[437.22, 267.32], [355.93, 266.35], [430.66, 244.38], [493.3, 244.95], [437.22, 191.73], [355.93, 191.54], [430.66, 187.15], [493.3, 187.26]]}, {\"category\": \"car\", \"corners_3d\": [[584.53, 217.75], [546.04, 217.51], [560.26, 212.82], [594.76, 213.01], [584.53, 184.37], [546.04, 184.3], [560.26, 183.1], [594.76, 183.15]]}, {\"category\": \"car\", \"corners_3d\": [[659.53, 199.55], [638.48, 199.28], [652.38, 197.65], [672.25, 197.89], [659.53, 180.23], [638.48, 180.15], [652.38, 179.7], [672.25, 179.77]]}, {\"category\": \"cyclist\", \"corners_3d\": [[683.96, 233.62], [707.7, 233.78], [711.42, 239.48], [685.45, 239.29], [683.96, 172.84], [707.7, 172.84], [711.42, 172.84], [685.45, 172.84]]}]\n```" - } - ] - }, - { - "id": 649, - "data_source": "Mono3DRefer", - "scene_name": "007303", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[23.89, 545.87], [-345.53, 546.02], [184.95, 338.63], [349.01, 338.6], [23.89, 201.17], [-345.53, 201.18], [184.95, 185.44], [349.01, 185.44]]}, {\"category\": \"car\", \"corners_3d\": [[676.51, 196.62], [709.6, 196.83], [706.38, 198.79], [670.64, 198.55], [676.51, 163.59], [709.6, 163.51], [706.38, 162.74], [670.64, 162.84]]}, {\"category\": \"car\", \"corners_3d\": [[635.41, 190.11], [608.51, 189.96], [622.81, 188.52], [647.53, 188.64], [635.41, 166.6], [608.51, 166.65], [622.81, 167.18], [647.53, 167.13]]}]\n```" - } - ] - }, - { - "id": 650, - "data_source": "Mono3DRefer", - "scene_name": "007304", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[654.75, 219.97], [688.56, 219.95], [701.29, 226.95], [662.46, 226.97], [654.75, 189.77], [688.56, 189.76], [701.29, 192.27], [662.46, 192.28]]}]\n```" - } - ] - }, - { - "id": 651, - "data_source": "Mono3DRefer", - "scene_name": "007402", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"van\", \"corners_3d\": [[644.7, 193.95], [666.85, 193.97], [670.19, 195.99], [645.92, 195.96], [644.7, 167.38], [666.85, 167.37], [670.19, 166.85], [645.92, 166.86]]}]\n```" - } - ] - }, - { - "id": 652, - "data_source": "Mono3DRefer", - "scene_name": "007402", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[719.04, 220.62], [676.77, 220.57], [669.65, 214.2], [706.27, 214.24], [719.04, 183.99], [676.77, 183.98], [669.65, 182.5], [706.27, 182.51]]}, {\"category\": \"car\", \"corners_3d\": [[652.64, 198.48], [674.66, 198.51], [678.11, 200.57], [654.32, 200.54], [652.64, 177.22], [674.66, 177.22], [678.11, 177.57], [654.32, 177.57]]}, {\"category\": \"car\", \"corners_3d\": [[3465.11, 1500.95], [2240.88, 1480.94], [939.17, 432.88], [1175.67, 433.67], [3465.11, 271.26], [2240.88, 269.78], [939.17, 192.12], [1175.67, 192.18]]}, {\"category\": \"car\", \"corners_3d\": [[400.35, 337.22], [240.29, 335.4], [391.61, 274.64], [492.48, 275.35], [400.35, 194.48], [240.29, 194.24], [391.61, 186.24], [492.48, 186.34]]}, {\"category\": \"car\", \"corners_3d\": [[740.51, 282.36], [839.06, 281.92], [953.01, 331.52], [810.05, 332.45], [740.51, 190.36], [839.06, 190.29], [953.01, 198.21], [810.05, 198.36]]}, {\"category\": \"car\", \"corners_3d\": [[485.39, 259.44], [399.05, 259.13], [447.44, 241.1], [515.83, 241.29], [485.39, 178.54], [399.05, 178.52], [447.44, 177.34], [515.83, 177.35]]}, {\"category\": \"car\", \"corners_3d\": [[814.46, 271.95], [722.49, 271.83], [695.01, 246.24], [763.14, 246.3], [814.46, 191.31], [722.49, 191.29], [695.01, 186.52], [763.14, 186.54]]}, {\"category\": \"car\", \"corners_3d\": [[540.55, 224.49], [492.63, 224.53], [506.23, 218.09], [548.18, 218.06], [540.55, 178.23], [492.63, 178.23], [506.23, 177.56], [548.18, 177.56]]}]\n```" - } - ] - }, - { - "id": 653, - "data_source": "Mono3DRefer", - "scene_name": "007402", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (van, car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[719.04, 220.62], [676.77, 220.57], [669.65, 214.2], [706.27, 214.24], [719.04, 183.99], [676.77, 183.98], [669.65, 182.5], [706.27, 182.51]]}, {\"category\": \"car\", \"corners_3d\": [[652.64, 198.48], [674.66, 198.51], [678.11, 200.57], [654.32, 200.54], [652.64, 177.22], [674.66, 177.22], [678.11, 177.57], [654.32, 177.57]]}, {\"category\": \"van\", \"corners_3d\": [[644.7, 193.95], [666.85, 193.97], [670.19, 195.99], [645.92, 195.96], [644.7, 167.38], [666.85, 167.37], [670.19, 166.85], [645.92, 166.86]]}, {\"category\": \"car\", \"corners_3d\": [[3465.11, 1500.95], [2240.88, 1480.94], [939.17, 432.88], [1175.67, 433.67], [3465.11, 271.26], [2240.88, 269.78], [939.17, 192.12], [1175.67, 192.18]]}, {\"category\": \"car\", \"corners_3d\": [[400.35, 337.22], [240.29, 335.4], [391.61, 274.64], [492.48, 275.35], [400.35, 194.48], [240.29, 194.24], [391.61, 186.24], [492.48, 186.34]]}, {\"category\": \"car\", \"corners_3d\": [[740.51, 282.36], [839.06, 281.92], [953.01, 331.52], [810.05, 332.45], [740.51, 190.36], [839.06, 190.29], [953.01, 198.21], [810.05, 198.36]]}, {\"category\": \"car\", \"corners_3d\": [[485.39, 259.44], [399.05, 259.13], [447.44, 241.1], [515.83, 241.29], [485.39, 178.54], [399.05, 178.52], [447.44, 177.34], [515.83, 177.35]]}, {\"category\": \"car\", \"corners_3d\": [[814.46, 271.95], [722.49, 271.83], [695.01, 246.24], [763.14, 246.3], [814.46, 191.31], [722.49, 191.29], [695.01, 186.52], [763.14, 186.54]]}, {\"category\": \"car\", \"corners_3d\": [[540.55, 224.49], [492.63, 224.53], [506.23, 218.09], [548.18, 218.06], [540.55, 178.23], [492.63, 178.23], [506.23, 177.56], [548.18, 177.56]]}]\n```" - } - ] - }, - { - "id": 654, - "data_source": "Mono3DRefer", - "scene_name": "007436", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (pedestrian). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (707.05, 707.05), (cx, cy) = (604.08, 180.51). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"pedestrian\", \"corners_3d\": [[214.37, 298.4], [174.13, 307.06], [106.07, 303.91], [149.92, 295.66], [214.37, 151.25], [174.13, 149.1], [106.07, 149.89], [149.92, 151.93]]}]\n```" - } - ] - }, - { - "id": 655, - "data_source": "Mono3DRefer", - "scene_name": "007462", - "question_type": "mobj_3dBbox_json", - "conversations": [ - { - "from": "human", - "value": "Please identify all target objects in this scene from the following categories: (car). Ignore objects that are too small or too far away to be recognized. Given the camera intrinsic parameters: (fx, fy) = (721.54, 721.54), (cx, cy) = (609.56, 172.85). Based on the provided information, and answer the 3D attributes of all target objects with JSON format including: \n- corners_3d: the projected points of 8 corners of the 3D bounding box in the image coordinate (list of [x, y] with unit in pixels).\n\nNoting the response JSON format should be like:\n```JSON\n[{\"category\": category, \"corners_3d\": [[x1, y1], [x2, y2], ..., [x8, y8]]}, ...]\n```" - }, - { - "from": "gpt", - "value": "```JSON\n[{\"category\": \"car\", \"corners_3d\": [[585.04, 202.32], [562.27, 202.23], [568.93, 200.48], [590.36, 200.56], [585.04, 181.47], [562.27, 181.45], [568.93, 180.94], [590.36, 180.96]]}, {\"category\": \"car\", \"corners_3d\": [[505.04, 233.35], [453.62, 233.14], [480.4, 224.71], [524.69, 224.86], [505.04, 187.26], [453.62, 187.21], [480.4, 185.21], [524.69, 185.24]]}, {\"category\": \"car\", \"corners_3d\": [[820.5, 285.17], [930.35, 285.69], [1091.73, 346.68], [921.77, 345.44], [820.5, 193.08], [930.35, 193.17], [1091.73, 204.15], [921.77, 203.93]]}, {\"category\": \"car\", \"corners_3d\": [[753.86, 250.07], [823.16, 249.77], [884.91, 269.41], [798.13, 269.87], [753.86, 179.9], [823.16, 179.87], [884.91, 181.66], [798.13, 181.71]]}, {\"category\": \"car\", \"corners_3d\": [[725.53, 231.39], [778.91, 231.56], [807.56, 243.59], [743.18, 243.33], [725.53, 182.77], [778.91, 182.8], [807.56, 184.83], [743.18, 184.79]]}, {\"category\": \"car\", \"corners_3d\": [[532.46, 220.93], [490.75, 220.74], [510.22, 215.22], [547.18, 215.37], [532.46, 184.93], [490.75, 184.88], [510.22, 183.49], [547.18, 183.53]]}, {\"category\": \"car\", \"corners_3d\": [[706.99, 221.24], [750.84, 221.33], [768.61, 228.56], [718.19, 228.44], [706.99, 181.62], [750.84, 181.64], [768.61, 182.95], [718.19, 182.92]]}, {\"category\": \"car\", \"corners_3d\": [[549.91, 214.42], [514.37, 214.32], [529.18, 209.62], [560.72, 209.7], [549.91, 185.0], [514.37, 184.97], [529.18, 183.6], [560.72, 183.62]]}, {\"category\": \"car\", \"corners_3d\": [[694.15, 212.54], [730.57, 212.64], [742.8, 218.41], [701.07, 218.28], [694.15, 179.46], [730.57, 179.48], [742.8, 180.44], [701.07, 180.42]]}]\n```" - } - ] - } -] \ No newline at end of file